KERMIT 95 BUG LIST
The Kermit Project, Columbia University
http://www.columbia.edu/kermit/
7 June 2002
For: Kermit 95 2.0
This file lists bugs or other problems in Kermit 95 and its operating
environment. There is a lot of information here, organized chronologically,
and so it is recommended you use a text editor or Web browser to do
case-insensitive searches for keywords to turn up what you are looking for.
Bugs that have been fixed are listed here by number and name, but, except for
the bugs that were fixed since version 1.1.16, the descriptions of fixed
bugs have been moved to a separate file.
If you do not have the current version of Kermit 95, please upgrade to it:
http://www.columbia.edu/kermit/k95patch.html
if you are experiencing bugs or problems, they might already be fixed. Also,
the BUGS.TXT / BUGS.HTM file that comes with the current release will have a
more up-to-date list of bugs and hints than an older copy. If you do have the
current version of K95, also consult:
- The Kermit 95 FAQ (Frequently Asked Questions).
- The
newbugs.txt
file on the Kermit website, which lists bugs discovered since the last release
of K95.
In this document, except where otherwise noted, the term Windows 95 or
Windows 9x refers to Windows 95, Windows 98, or Windows ME,
but not to Windows NT, 2000, or XP. Similarly, the term
Windows NT also includes Windows 2000 and XP, but not Windows 95,
98, or ME.
[ Index ] [ Bug Descriptions ] [ K95 Home ] [ Kermit Home ]
- BUG CATEGORY
- (M) Microsoft Windows bug.
(I) An IBM OS/2 bug.
(X) A feature that is disabled in Kermit 95 because of
an operating-system bug.
(Z) A bug in the Dialer caused by a bug in the GUI
development system that was used to build it.
(V) Bug or restriction in vendor code, e.g. in 3rd-party
TCP/IP stacks.
(K) A Kermit 95 limitation or bug.
(D) A documentation error.
(C) A hardware / System / Drivers Configuration problem.
( ) Not a bug.
NOTE: The February 1996 Service Pack from Microsoft did NOT fix any of
the bugs in the M and X categories. The September 1996 OEM SR2 release
fixed at least one of these bugs
(Bug #79) but is not available as a patch
to earlier releases.
IMPORTANT: Windows NT 4.0 for Intel shipped with a dangerous but
little-known bug that can randomly corrupt your data. Microsoft has fixed the
problem in Service Pack 1 for Windows NT 4.0. It is strongly recommended
that users of Windows NT 4.0 download and install this and all later fixes.
- BUG STATUS
-
[ ] Indicates an item that is still open or current
[+] Indicates the bug is fixed or worked around as of the current release
[-] Indicates bug is partially fixed or a workaround is given
[*] Information and/or not really a bug
- [-] (M) Ctrl-C Interruption in Command Screen
- [+] (X) Video Attributes vs Terminal Screen Size in Windows 95
- [+] (K) Microsoft Telephony (TAPI)
- [*] (K) Registration
- [ ] (M) Changing Code Pages
- [-] (M) Keyboard Scan Codes, Caps Lock, and Dead Keys
- [-] (D) "Kermit 95" Keyboard Verb table
- [ ] (M) PUSH and related commands do not work in Windows NT 3.50
- [-] (M) Don't Use the Auto Font Setting in Windows 95
- [+] (Z) Blotchy Dialer Fields
- [ ] (K) Missing Features
- [*] (C) Hardware / system / drivers configuration
- [-] ( ) Miscellaneous Dialer peculiarities
- [*] ( ) Miscellaneous Hints and Tips
- [*] ( ) Debugging Tips
----(Version 1.1.1)----
- [+] (Z,M) K-95 Dialer Vertical Scrolling
- [+] (K) ZMODEM Downloads
- [+] (K) ZMODEM and TELNET
- [+] (K) Alphabetization of Dialer Entries
- [+] (K) K95 Sometimes Thinks a Connection is Open When It Starts Up
- [+] (K) Download Directory Doesn't Work
- [+] (K) SET MODEM commands in K95CUSTOM.INI have no effect
- [+] (K) Kermit 95 sometimes can't find its script or key mapping files
- [+] (K) The Dial Prefix also becomes the Dial Suffix
- [+] (K) Dialer complains unnecesarily about range of certain numbers
- [+] (K) Kermit BBS entry in Dialer has wrong backspace key setting
----(Version 1.1.2)----
- [*] (C) Screen updates updates are slow on some PCs
- [+] (K) ZMODEM Recovery
- [+] (K) ANSWER command crashed Kermit 95
- [+] (K) Kermit always turned off autoanswer in the modem
- [+] (K) Compose Key problems
- [+] (K) SET TERMINAL CHARACTER-SET TRANSPARENT did not show all characters
- [+] (K) Various other VT emulation problems
- [+] (K) Printer problems
- [+] (K) \v(keyboard) variable improperly set
- [+] (K) XMODEM file transfer problems
- [+] (K) MOVE and MMOVE command did not work right with XYZMODEM
- [+] (K) Ctrl-C interruption nonfunctional in certain cases
- [+] (K) File transfer recovery after Ctrl-C
- [+] (K) SET FILE DOWNLOAD-DIRECTORY didn't work with server mode
- [+] (K) Filenames could not be parsed from variables
- [+] (K) Choosing terminal character-set in the Dialer
- [+] (K) The EMACS.INI file was not properly patched in version 1.1.1
- [+] (K) The 1.1.1 patch versus READ.ME files in multiple directories
- [+] (M) Xon/Xoff flow control doesn't work in Windows 95
- [-] (M) Server mode sometimes locks up in Windows 95
- [-] (M,K) Kermit 95 file naming conventions in conflict with Windows
- [+] (K) The Dialer Login page does not show LOGIN.SCR in the text box
- [ ] (K) The K-95 icon does not show up in Windows NT 3.5x
- [+] (K) SEND and RECEIVE as-names ignored by XYZMODEM
- [*] (D) Missing files
- [+] (K) Title bar not properly set from "kermit -j host"
----(Version 1.1.3)----
- [ ] (M) Can't interrupt SET HOST or TELNET or RLOGIN command
- [+] (K) SET FILE DOWNLOAD-DIRECTORY does not work with XYZMODEM
- [+] (K) Crash when pasting voluminous material into terminal window
- [+] (K) Problem with SET FILE DOWNLOAD-DIRECTORY<Enter>
- [+] (K) Terminal reset kills K95 if TERM CHARACTER-SET is TRANSPARENT
- [+] (K) Cursor-type selection made in Dialer not saved
- [+] (K) PBX-related entries in the Dialer's Dialing Location page not saved
- [+] (K) No choice for location of Kermit 95's temporary files
- [ ] (M) The "Paste" button on the Terminal Windows Toolbar doesn't work.
- [+] (K) EMACS.INI Alt-x = Meta-x doesn't work
- [+] (K) INPUT could not be canceled from the keyboard
- [+] (K) SET ALARM / IF ALARM doesn't always work
- [+] (K) \v(cpu) always says "unknown".
- [+] (K) Invalid escape sequence ESC [ ? 6/7 i caused problems
- [+] (K) Problems with transparent printing and accented letters
- [+] (K) Microcom Deskporte modem configuration wrong
- [+] (K) Cursor-type selection not saved by Dialer
- [+] (K) PBX material from Locations & Codes Dialer page not saved
- [+] (K) SET TERM CHAR TRANSPARENT, then terminal reset, causes crash
- [+] (K) Spurious "Connection might still be active on COM1..." messages
- [+] (K) SET PORT<Enter> gives spurious error message
- [+] (K) DIAL timeouts sometimes caused an "access violation" crash
- [+] (K) Modem list in SETUP program PPI / Multitech mixup
- [+] (K) Incorrect default assignment for Ctrl-Space key combination
- [+] (K) "k95 filename" requires fully specified filename
- [+] (K) XYZMODEM Transfers didn't work over RLOGIN connections
- [+] (M) PCMCIA or Plug-and-Play Modem Problems
- [ ] (V) RLOGIN vs FTP Software OnNet-32
- [ ] (K,V) "set host 0" vs FTP Software OnNet-32
- [*] (V) Kermit 95 and Digital UNIX 3.2
- [ ] (M) Extraneous characters on terminal screen in Windows 95
- [+] (M) Special keys on national keyboards don't work
- [+] (K) VT220/320 Erase Character (ECH) bug
- [*] (K) Intel Modems without Fax
- [+] (K) LOGIN.SCR renamed to LOGIN.KSC to avoid conflicts with screen savers
- [?] (V) Uploads on LAT connections tend to fail
- [+] (K) XYZMODEM and Kermit file transfer screens and stats are inconsistent
- [ ] (M) Printing to network printers in Windows 95 or NT might not work
----(Version 1.1.4)----
- [+] (K) Messed up colors on initial Terminal screen
- [+] (K) Colors behave differently than in MS-DOS Kermit
- [+] (K) Printing to PostScript printers doesn't work
- [+] (K) Redialing on a SET HOST / SET MODEM / DIAL connection didn't work
- [+] (K) Command-file line numbers not reset
- [+] (K) ASSIGN/DEFINE failed to fail if definition too long
- [+] (K) Dialer temporary files called .SCR instead of .KSC
- [+] (K) The MSEND command did not work with ZMODEM
- [+] (K) Backspace key didn't work with "set term char finnish"
----(Version 1.1.5)----
- [ ] (K) Kermit 95 requires long filenames
- [-] (K) Host-controlled status lines
- [+] (K) Wyse screen-transmit functions not implemented
- [+] (M) Windows 95 Dead key and Caps Lock bugs are fixed! (mostly)
- [*] (K,M) East European keyboard features might not work in Windows 95
- [+] (K) VT emulation DCS/OSC/APC/etc bugs
- [+] (K) Terminal color bugs in version 1.1.4
- [ ] (K) VT220/320 downloadable fonts / soft character sets don't work
- [+] (K) SET TERMINAL CHARACTER-SET PORTUGUESE didn't work
- [+] (K) VT emulation and SET TERMINAL CHARACTER-SET FINNISH (etc) bug
- [+] (K) Autodownload versus SET TERM CHARACTER-SET
- [+] (M) Keyboard doesn't work if I start K95.EXE from Netscape
- [+] (Z) Dialer radio buttons not lined up
- [+] (K) Multiple sessions to same Dialer entry don't work
- [+] (M,K,Z) Can't select a Dialer entry from the keyboard
- [+] (K) Dialer crash after editing an entry entry name
- [+] (K) Dialer PBX configuration problems
- [ ] (M) Ctrl-C doesn't work in NT when K95 started from the Dialer
- [+] (K) No way to make LAT connections from the Dialer
- [*] (M) Kermit 95 and Internet firewalls, SOCKS, etc
- [+] (K) Unreasonable delays in SET HOST / TELNET to numeric IP address
- [ ] (V) Kermit 95 does not work with LWP 5.0 TCP/IP in Windows 95
- [ ] (M) No way to select split-speed serial communication
- [+] (K) Errors in built-in modem support
- [+] (K) SEND did not work with filename completion
- [+] (K) OUTPUT could not send backslash characters
- [+] (K) OUTPUT could not send long strings
- [+] (K) Extraneous material in session log during INPUT
- [+] (K) INPUT buffer too short
- [+] (K) SEND foo\%a did not work
- [+] (K) No way to print on A4 paper using textps
- [+] (K) Invalid Page Fault after using command recall a lot
- [ ] (M) Alt- doesn't work.
- [-] (K) Interruption keys are ignored during XYZMODEM transfers
----(Version 1.1.6)----
- [*] (C) Tone dialing changes to Pulse after first digit
- [+] (K) VT100 reverse scrolling broken
- [+] (K) SET KEY commands became slow
- [*] (C) Terminal bell doesn't work as before, or at all
- [+] (K) SET MSKERMIT KEYCODES ON did not work with some gray keys
- [+] (K) Array elements could not be used for file names
- [+] (K) Keyboard modes can't be modifed
- [-] (M,K) Control keys on national keyboards in Windows 95
- [+] (K) SCOANSI and AT386 emulation vs Code Page 437
- [+] (K) SCOANSI and AT386 emulation vs CSI 10/11/12 m
- [+] (K) Bad characters in Registration name
- [+] (K) SET MODEM commands in K95CUSTOM.INI are ignored by Dialer
- [ ] (K) Completion does not work with directory names
- [ ] (K) Command-line editing doesn't use PC editing keys
- [ ] (K) The DIRECTORY and TYPE commands are too slow
- [+] (K) DIRECTORY command gives spurious error message
- [+] (K) XYZMODEM transfer statistics are off
- [+] (K) SET TERMINAL ESCAPE-CHARACTER DISABLED didn't work
- [ ] (K,M) Command screen messed up after Ctrl-C
- [+] (K) Cursor disappears after Ctrl-C interruption of file transfer
- [+] (K) \v(sysid) was garbage
- [+] (K) SHOW KEY ALL did not show keys that had string definitions
- [+] (K) Dialer File Transfer page items spuriously grayed out
- [+] (K) SET TERM COLOR ERASE CURRENT-COLOR didn't work
- [ ] (K) Patching up from version 1.1.3 might not work
- [+] (K) XYZMODEM file transfers gave incorrect status codes
- [+] (K) XYZMODEM transfers made no transaction log entries
- [+] (K) TELNET negotiations were sometimes skipped
- [+] (K) Data loss during file transfer
- [+] (K) K95 could not send a file that was already open
- [+] (K) Dynamic packet timeout values were sometimes too high
- [+] (K) Character attributes not shown in Televideo emulation
- [ ] (V) PATCH.EXE only works with FAT filenames
- [+] (K) TEXTPS.EXE was broken in version 1.1.5
- [+] (K) File transfer efficiency number is misleading
----(Version 1.1.7)----
- [+] (K) VT220/320 DCS string cancellation didn't work
- [+] (K) SCOANSI and AT386 emulation conflict with Dialer
- [+] (K) Server GET fails if a SEND-LIST had previously been used
- [+] (K) PUSH from Terminal screen to shell could not be returned from
- [+] (Z) K95DIAL window minimizes to desktop instead of task bar
- [*] (K) Referring to directory names that start with digits
- [+] (K) Televideo and Wyse printing didn't work right
- [+] (K) Wrong default character-set for VT terminals in Dialer
----(Version 1.1.8)----
- [ ] (M) The Dialer database can be corrupted by other applications
- [-] (K) The Dialer can seem to disappear
- [+] (K) The "Chess" entry in the Dialer is wrong
- [*] (K) XMODEM-CRC, YMODEM-CHECKSUM, and ZMODEM-CRC16 are not supported
- [+] (K) Renaming downloaded files with XYZMODEM doesn't work
- [+] (K) XYZMODEM downloads are somewhat slower than expected
- [+] (K) Scroll Lock doesn't work in version 1.1.7
- [+] (K) No way to change the default key map for each terminal type
- [+] (U) CTRL2CAP instructions misleading
- [+] (K) LOG SESSION erroneously translates character sets
- [+] (K) VT102 insert character function doesn't work
- [+] (K) Spurious keyboard locks in VT220/320 emulation
- [+] (K) Improper handling of Alternate ROM in VT emulation
- [+] (K) Erroneous VT emulator Secondary and Tertiary DA Reports
- [+] (K) Erroneous DECID reports in VT220/320 emulation
- [+] (K) VT emulation Horizontal Position Absolute (HPA) ignored
- [+] (K,C) Failure of K95 to start on certain Windows NT systems
- [+] (K) Simultaneous LAT sessions don't work
- [ ] (X) No redirection of standard input
- [+] (K) SHOW KEY does not properly display C0 and 8bit characters
- [+] (K) "set command more-prompting" could not be parsed
- [+] (K) No way to specify 0 or no timeout in INPUT
- [+] (K) INPUT ignores autodownload sequences
- [+] (K) Problems entering commands that end with backslash
- [+] (K) Spurious invisible Z command that does nothing
- [ ] (M) Close box in K95 title bar doesn't work in Windows 95
- [*] (K) Dialer startup is slow
- [*] (K) Dialer toolbar buttons might "wrap around"
- [+] (K) Statusline OFF vs TELNET NAWS bug
----(Version 1.1.9)----
- [+] (K) Page faults in K95 version 1.1.8
- [+] (K) Keyboard translations don't work in Windows NT in K95 1.1.8
- [+] (K) Printing did not work right in version 1.1.8
- [+] (K) Dead-key and Alt-Gr broken in 1.1.8
- [+] (K) APC during INPUT broken in 1.1.8
- [+] (K) Redundant terminal-type change in 1.1.8
- [ ] (K) CD to nonexistent directory can cause errors
- [+] (V) Problems with host-based WordPerfect
- [+] (K) Russian and Hebrew keyboard modes don't work
- [*] (M) Why does Dialup networking pop up whenever I make a Telnet connection?
- [*] (K) Why does the termimal status line say "Script:" instead of "Prompt:"?
- [*] ( ) Why is my login prompt missing?
- [+] (K) VERSION command did not display copyright notice
- [+] (K) Alt-: and Alt-; missing from EMACS mode
- [+] (K) Dialer would not start if any database entries were corrupt
- [+] (K) Russian and Hebrew keyboard mode conversions broken
- [+] (K) Latin-2 Compose sequences broken in 1.1.8
- [*] (M) Windows 95 QuickEdit interferes with Kermit 95's mouse functions
- [ ] (M) Caps Lock state can change when moving between windows
- [+] (K) LOGIN.KSC broken in version 1.1.8
- [+] (K) ADD SEND-LIST followed by MOVE did not delete source files
- [+] (K) Marking text on bottom line of terminal screen didn't work
- [+] (K) Sending data from terminal screen when SEND-DATA OFF
- [+] (K) Kermit file-transfer packets sent out of window
- [+] (K) VMS SET TERM/WIDTH did not clear the screen
- [+] (K) k95 "file name with spaces" did not work in 1.1.8
- [+] (K) k95 command screen could be frozen or corrupted in 1.1.8
----(Version 1.1.10)----
- [+] (K) Spurious cursor in 1.1.9
- [*] (M) Appearance of horizontal scrolling indicators
- [*] (K) Network editing hints
- [+] (M) Windows 95 long screens don't work in 1.1.9
- [+] (V) Problems with TES32
- [ ] (M) Keystrokes might be transmitted out of order in Windows 95
- [ ] (V) Conflicts with CompuServe, AOL, etc
- [+] (K) HPTERM and HP2621 screens were fractured
- [+] (K) Overzealous recording in session log
- [+] (K) VT100/102 keypad not mapped to PC keypad by default
- [+] (K) Assorted window sizing problems in 1.1.9
- [+] (K) SET TERMINAL CURSOR did not enable the cursor
- [+] (K) Autoupload problems with XMODEM and YMODEM (K)
- [+] (K) Data General DASHER terminal emulation problems
- [+] (K) Default editor set in K95.INI might be wrong
- [+] (K) Dial-in host mode did not agree with user-defined modems
- [ ] (K) Not all SET TERMINAL settings are shown by SHOW TERMINAL
- [ ] (C) PUSH, RUN, and similar commands might not work
- [+] (K) Unwanted alert box from K95CINIT
- [+] (K) Problems when switching from VT to ANSI terminal type
- [*] (V) HINT: Using SCOANSI emulation with SCO hosts
- [ ] (M) Environment variables are not case sensitive
- [+] (M) Data loss in dialup host mode
- [+] (K) Problems with automatic login from certain Dialer entries
- [+] (K) Problems with arrow keys and command recall
- [+] (K) Misleading Dialer error message about screen dimensions
- [*] (K) HINT: How to disable scrollback
- [+] (K) Wildcard sends don't work right in XYZMODEM
- [*] (K) EXPLANATION: Changes in the operation of the numeric keypad
- [+] (K) DEC NRCs are forgotten in Multinational Mode
- [+] (K) ZMODEM crash recovery versus some FILE COLLISION settings
- [+] (K) ZMODEM autodownload doesn't work with Wyse, VC, or DG
- [*] (C) HINT: Kermit mouse buttons versus nonstandard mouse setups
- [+] (K) VT terminal emulation wrapping problem
- [+] (K) WP51.INI (WordPerfect 5.1 key mapping file) had a typo
- [*] (K) HINT: Using the NETEDIT macro with GUI editors
- [ ] (M) Yellow background change after !edit in K95
- [+] (K) Popup printer-error boxes causes Kermit 95 to hang
- [+] (K) Host-directed printing did not work in DG emulation
- [+] (K) INPUT/MINPUT did not handle Telnet negotiations correctly
- [+] (K) Dialer Options..Dialing page PBX values handled incorrectly
- [+] (K) Wyse character graphics bug
- [+] (K) Transparent printing of binary files did not work
- [+] (K) VT100 terminal ID was incomplete
- [ ] (K) Kermit/2 keycodes are different from OS/2 C-Kermit 5A(191)
- [ ] (C) Kermit/2 conflicts with SWAPDCP on OS/2
----(Version 1.1.11)----
- [*] (M) RLOGIN versus Microsoft's "WINNUKE" Patches
- [+] (K) DIALINF.DAT can't be read-only
- [ ] (M) Caps Lock state can be lost when switching between windows
- [+] (K) Serial-port file transfer fails with large packet x window size
- [ ] (M) Can't change keyboard language in Windows 95
- [+] (K) K95 does not delete startup file generated by the Dialer (K)
- [+] (K) Dialer can leave temporary files behind (K)
- [+] (K) Kermit 95 TELNET.EXE stub works only if K95 directory in PATH
- [+] (K) Problems with incoming TCP connections in OS/2
- [+] (K) Dialer could crash if an entry's view status changed
- [+] (K) SHOW KEY did not always give accurate results
- [+] (K) REDIRECT command didn't work
- [+] (K) Incoming NUL (0) characters confuse INPUT command
- [+] (K) RECEIVE as-name did not work with XYZMODEM
- [+] (K) Various EMACS mode bugs
- [+] (K) Kermit protocol might send packet out of window
- [+] (K) MOVE from SEND-LIST did not delete original files
- [+] (K) Application filename could not contain spaces
- [+] (K) Hayes high-speed-modem init string problem
- [+] (K) READ command does not fail if file not open
- [+] (K) Problems with long function arguments
- [+] (K) Certain \function()s can misbehave
- [+] (K) X MOD 0 crashes program
- [+] (K) ANSWER doesn't automatically CONNECT
- [+] (K) Overzealous EXIT warning
- [+] (K) OUTPUT doesn't echo when DUPLEX is HALF
- [+] (K) Minor problems with REMOTE DIRECTORY/DELETE/etc
- [+] (K) CHECK command broken
- [+] (K) Problem with SET TRANSMIT ECHO
- [+] (K) Dialing results were sometimes incorrect
- [+] (K) IF EXIST and IF DIRECTORY versus spaces in filenames
- [+] (K) CD, SET FILE DOWNLOAD-DIRECTORY, etc, don't provide full service
- [+] (K) READ command too picky about last or only line in file
- [*] (K) No way to include braces in function arguments
- [+] (K) DIALINF.DAT file opened in READ/WRITE mode
- [+] (K) Missing and incorrect key definitions in EMACS mode
- [+] (K) NULs were being stripping during file transfers and printing
- [+] (K) XYZMODEM RECEIVE was ignored
- [+] (K) SET STARTUP-FILE DISCARD did not always delete the file
- [+] (K) SHOW MODEM didn't quote backslashes
- [+] (K) WIN32 COMM TX BUFFER SIZE set too small
- [+] (K) K2DC.EXE accidently closed socket of incoming connection
- [+] (K) Auto-print Formfeed bug
- [+] (K) SAVE KEYMAP and SHOW KEY replaced RS with ESC
- [*] (M) How can I use SSH, SSL, CHAP, etc, to make secure connections?
- [*] (M) Why can't I pick any Windows font I want?
- [+] (V) Data loss on Meridian Technology SuperLAT connections
- [*] (K) Hints on dialing toll-free numbers
- [*] (K) Hints on "blind dialing"
- [*] (K) Log files in multiple sessions
- [+] (K) Hazeltine 1500 emulation did not work at all
- [+] (K) Problems with Wyse 50 emulation
- [+] (K) Problems in DG DASHER 217 emulation
- [+] (K) HPTERM emulation did not handle line-drawing & math
- [+] (K) Many Dialer entries were obsolete
- [+] (K) Unwanted triangles in Wyse, TVI, and Hazeltine emulation
- [+] (K) Spurious spaces in ANSI emulations
- [+] (K) Data loss/corruption on Windows 95 serial connections
----(Version 1.1.12)----
- [ ] (K) Commands like "RUN D:\FOO\BAR\BAZ.BAT" don't work in Windows
- [ ] (M) "RUN /K...." hangs K95 in Windows
- [+] (K) Commands like "SEND \\K95\\TMP\\OOFA.TXT" did not work
- [+] (K,M) Problems with help and completion in filenames with \\
- [ ] (M) CTL3D32.DLL is missing
- [ ] (M) K95 might take a long time to start and stop
- [+] (M) K95 might not be able to exit on some Windows 95 systems
- [+] (K) Dialer might crash if no modem was defined
- [+] (K) Memory leak in Dialer
- [+] (K) URL hot spot could fail if URL at end of sentence
- [+] (K) EDIT and NETEDIT might have stopped working in 1.1.12
- [+] (K) Telnet negotiations could become frozen
- [+] (K) K95 could become stuck in marking mode
- [+] (K) HP terminal emulation problems in 1.1.12
- [+] (K) Wyse terminal emulation problems in 1.1.12
- [+] (K) K95 could hang upon screen-size change request
- [+] (K) SET MOUSE ACTIVATE command broken in 1.1.12
- [+] (K) Character set versus copy/paste from/to Clipboard
- [+] (K) (void)
- [ ] (M) Paste to host using Toolbar Paste sends Esc O P when Num Lock on
- [ ] (K) Reverse screen-writing direction not supported
- [+] (K) Transparent print problems using 8-bit CSI
- [+] (K) APC did not block terminal emulator
- [+] (K) Conflict between INPUT and APC
----(Version 1.1.13)----
- [+] (V) K95 might not work with certain third-party TCP/IP stacks
- [+] (K) Another problem with filenames containing spaces
- [+] (K) CONNECT mode hangs when host drops connection or \Kquit is used
- [+] (K) Problem with Num Lock and SET MSKERMIT KEYCODES ON
- [+] (K) Problem with OUTPUT command when there is no connection
- [+] (K) END from within SWITCH doesn't work
- [ ] (V) URL hot spots might not work
- [+] (K) (void)
- [+] (K) (void)
- [+] (K) (void)
- [+] (K) (void)
- [+] (K) (void)
- [+] (K) (void)
- [+] (K) (void)
- [+] (K) (void)
- [+] (K) (void)
- [+] (K) (void)
- [+] (K) (void)
- [+] (K) (void)
- [+] (K) (void)
- [+] (K) (void)
- [+] (K) (void)
- [+] (K) (void)
- [+] (K) (void)
- [+] (K) (void)
- [+] (M,K) Problems with Dead Keys and Alt-Gr keys in Windows 95
- [+] (K) Kermit vs TAPI Dialing in the Dialer
- [ ] (M) K95 vs Num Lock, Caps Lock, and Scroll Lock
- [*] (M) Parallel Port Communications
- [+] (K) Wrong system type in Telnet NEW-ENVIRONMENT negotiation
- [+] (K) IF ALARM might not work
- [+] (K) SET TERM REMOTE-CHAR LATIN6 doesn't work
- [+] (K) Single-click mouse actions generate an extraneous error beep
- [+] (K) Various terminal emulation bugs
- [+] (K,V) Ctrl-C caught by 4NT.EXE instead of K95 in Windows NT
- [+] (K) OS/2 K2DIAL loses data in the Login Script text box
- [+] (K) Various errors in the Windows Telephony (TAPI) implementation
- [+] (K) Dialing fails due to garbled modem command
- [+] (K) Dialer Modem Commands reset when modem selected for editing
- [+] (K) SHOW KEY interrupted by Ctrl-C would cause keyboard problems
- [+] (K) SET HOST * results in 100% CPU Utilization
- [+] (K) Telnet to non-23 port might fail to produce a login prompt
- [+] (K) SET MODEM TYPE to certain types didn't work
- [+] (K) SETUP.EXE would crash while displaying summary of choices
- [+] (K) Default AT386 and SCOANSI keymaps reference DEC Kverbs
- [+] (K) SET TERMINAL DEBUG ON interferes with TELNET negotiations
- [+] (K) OS/2 FTP PC/TCP stack would cause K2 to crash
- [+] (K) K95's -N command-line option in OS/2 would not accept parameters
- [+] (K) SET BROWSER ignored by \KmouseURL keyboard verb
- [+] (V) K95 could not be the Telnet client for Netscape Communicator
- [*] (K) The PRINT command no longer uses DOS COPY to print a file
- [+] (K,M) DIAL DISPLAY ON didn't work with TAPI modems
- [+] (K) TAP alpha paging script was not robust enough
- [+] (K) Incorrect setup of Telebit modems
- [+] (K) File transfers could fail when using very short packets
- [+] (K) Screen-printing problems in Windows NT
- [*] (V) Incorrect characters on screen when accessing SCO or AIX hosts
- [ ] (V) K95 is unable to print to WinPrinters
- [ ] (K) Host mode requires VT100 or ANSI client
- [+] (K) HPTERM and VIP7809 selection didn't work in the Dialer
- [+] (K) File-transfer problem using undefined variable for as-name
- [+] (K) Overzealous EXIT WARNING
- [+] (K) K95 program exit status codes were not always correct
- [+] (K) Various misleading file-transfer error messages
- [+] (K) EDIT command succeeds only on first attempt
- [ ] (M) "WINOLDAP - This program has performed an illegal operation..."
- [+] (K) Mixups in certain terminal-specific line-drawing character sets
- [+] (K) Unnecessarily difficult to get big window and packet sizes
- [+] (K) Host mode would not send file groups
- [-] (K) Problem using Swedish (etc) NRC and DEC Special Graphics
----(Version 1.1.14)----
- [+] (K) Various problems with the 1.1.14 Patch
- [+] (K) SET TELNET TERMINAL TYPE affects actual terminal type
- [+] (K) Problems with Kermit transfers on some Telnet connections
- [+] (K) AIX and HFT arrow keys didn't work in the Command screen
- [+] (K) AIX and HFT emulation versus Autowrap
----(Version 1.1.15)----
- [+] (M) Assorted problems caused by faulty optimization
- [-] (M) Problems with Winsock 2
- [+] (K,M) Redirection of Standard Output
- [*] (M,C) Illegal operations caused by lack of swap space
- [*] (M) Roaring Fan
- [+] (K) Failure to detect default printer in Windows NT
- [+] (K) Console mode changed from text to binary and not restored
- [-] (M) Trouble with EXIT under certain circumstances
- [+] (K) Overzealous EXIT warning (again)
- [+] (K) RUN command did not return status
- [+] (K) Problems with DIRECTORY command filename completion
- [+] (K) The DIRECTORY command would return one file too few
- [+] (K) MKDIR command did not work in OS/2
- [+] (K) SET TITLE effect was delayed
- [*] (M) BEEP command can be delayed
- [+] (K) CONNECT should not work if TAPI connection not open
- [+] (K) TAPI device not released when closed
- [+] (K) Various SET TCP command problems
- [+] (K) \v(ipaddr) didn't work before a connection was made
- [+] (K) Assorted Telnet initialization and negotiation problems
- [+] (K) Spurious NUL characters on Telnet connections
- [+] (K) Failure to recognize RLOGIN URLs
- [+] (K) Problems with Wyse graphics character sets in Windows NT
- [+] (K) QANSI graphics characters displayed as ASCII
- [+] (K) Problems with DG transparent printing
- [+] (K) Scrolling failures in ANSI X3.64 emulations
- [+] (K) Incorrect appearance of VT220/320 characters after erasure
- [+] (K) SCOANSI and AT386 function-key programming broken
- [+] (K) Various problems in HPTERM emulation
- [+] (K) Problems in Siemens/Nixdorf BA80 emulation
- [+] (K) Spurious red lines on debug-mode terminal screens
- [*] (K) Alt-x, Alt-h, etc, don't seem to work in AIXTERM
- [+] (K) Difficulties pasting
- [+] (K,M) Echoing delays
- [+] (K) MSKERMIT keycode for Ctrl-Enter was incorrect
- [+] (K) APC command containing SET TERM TYPE would stop
- [+] (K) Failure to receive files with long names on some systems
- [+] (K) Autodownload failures with parity mismatch
- [+] (K) File transfer thermometer might not complete
- [+] (K) XYZMODEM statistics counted each file twice
- [+] (K) SET TERMINAL TYPE might mishandle 7-bit NRCs
- [+] (K) Problems typing 7-bit national characters
- [+] (K) Problem Selecting HP-Roman8 character set from the Dialer
- [+] (K) RETURN within SWITCH didn't work
- [+] (K) \ffiles() counted directories when it shouldn't
- [+] (K) Problems with REMOTE QUERY KERMIT function()
- [-] (C) K95 Host Mode might be slow to put up Login prompt
- [*] (K) Host mode requires ANSI or VT terminal emulator
- [+] (K) Host-mode maximum users setting wouldn't stick
- [+] (K) Dialer Login Script text box data lost in OS/2
- [+] (K) Dialer Printer Type combo box could not be set
- [+] (K) Dialer Communications page speed entry wouldn't stick
- [*] (K) HINTS: Alphanumeric Paging
- [*] (K) HINTS: For blind users
- [*] (V) HINTS: Telnetting to OS/2
- [*] (V) HINTS: How to use K95 with Netscape Communicator 4.0
- [-] (M) TEMP or TMP environment variable definition can cause problems
- [+] (K) The GET Command Versus Filenames with Spaces
- [*] (K) HINT: CR and/or LF disappear!
- [*] (K) HINT: Eliminating "Press a key to continue" after RUN Command.
- [+] (K) Cursor positioning on double-width lines
- [+] (K) Errors in Default Keymap for EMACS Mode
- [+] (K) SET HOST * Port Not Freed
- [+] (K) Unwanted Command Screen Size Change
- [+] (K) MKDIR/RMDIR Problems with UNCs
- [+] (K) MKDIR/RMDIR Problems when Used in Macros or Command Files
- [+] (V) RUN, REDIRECT, PIPE Interaction with Nonstandard Shells
- [+] (K) Compression Skipped in Text-Mode Kermit File Transfers
- [+] (K) Slow Local Echo in Terminal Screen
- [+] (K) Slow Echo in Terminal Screen on Serial Connections
- [+] (K) Problems with Bidirectional Printer Configuration in Dialer
- [+] (K) Missing Error Messages in OUTPUT and INPUT
- [+] (K) Problems Editing Certain Commands
- [+] (K) Redirection of Certain REMOTE Commands Didn't work
- [+] (V) K95 did not work well with Citrix Winframe
- [+] (K,C) Wyse 370 colors didn't work
- [ ] (M) CJK Input Method Editors do not work in Windows 95
- [*] (V) Conflict between mouse and screen saver
- [+] (K) Improper bolding in VT reverse-screen mode
- [ ] (K) Dialer does not work in OS/2 2.x
- [ ] (M) Log files can disappear in Windows 95 if K95 killed
----(Version 1.1.16)----
- [ ] (V) Code Page 856 isn't really a code page
- [+] (D) SET TERMINAL IDLE-xxx command misdocumented
- [ ] (K) New screen updating method might not be good for everybody
- [+] (M,K) Problems invoking RLOGIN.EXE and TELNET.EXE
- [+] (K) AIXTERM, HFT, SCOANSI, AT386 scrolling problem
- [+] (K) Program crash during XMODEM transfer
- [+] (K) No means to receive XMODEM transfers with CRC block checking
- [+] (K) (Two bugs, One Number)
- [+] (K) Invalid TMP or TEMP directory will cause K95DIAL to crash
- [+] (K) INPUT command sets \v(instatus) to wrong value
- [+] (K) SET TERMINAL ESCAPE-CHARACTER does not work
- [+] (K) TCP/IP sessions do not work on OS/2 3.0
- [+] (K) OS variables not properly set in OS/2
- [ ] (M) Win95/98 does not randomly access pooled IP hosts
- [ ] (M) Kerberos authentication fails with "invalid network address"
- [+] (K) PAUSE command always reports FAILURE
- [+] (K,M,I) Kermit 95 could not display the EURO character
- [+] (K) TES32 for Win95 connections fail immediately
- [+] (K) \fsplit() function parses extra words
- [ ] (M) Cursor can disappear in command screen
- [+] (K) '\' at end of key definition cannot be sent to host
- [+] (K) LOOKUP command does not return failure
- [+] (K) Function rindex fails to match first instance of string
- [+] (K) Non-CP437 extended characters displayed improperly in NT
- [+] (K) In NT, dead keys produce an error beep when pressed
- [+] (K) SET BELL VISIBLE does not work reliably
- [+] (K) SET TELNET ENCRYPTION {REFUSED, ACCEPTED, ... } broken
- [+] (K) Max Command Length is not necessarily max length of macros
- [+] (K) END command numeric value is inaccessible to scripts
- [+] (K) 10-digit dialing rules forgets to dial the area code
- [+] (K) GET/REGET with SET FILE PATTERNS ON may corrupt files
- [+] (K) REMOTE commands may set STATUS to SUCCESS on FAILURE
- [+] (K) RUN command interferes with screen updates in a script
- [+] (K) \fword() fails to find the last word in a string
- [+] (K) Partial Dialing misuses dialing rules
----(Version 1.1.17)----
- [+] (K) The PRODUCT macro had an extraneous parameter
- [ ] (K) SETUP from CDROM might not exit cleanly
- [+] (K) URL hot spots vs Netscape Communicator
- [+] (K) Missing files (not really)
- [+] (K) DATE command does not return to command prompt
- [+] (K) CPS display in XYZmodem transfers is incorrect
- [ ] (K) Problems with SET FILE INCOMPLETE DISCARD
- [+] (K) Crash during Zmodem download with SET FILE COLLISION RENAME
- [+] (K) TAPI dial gives "invalid call state message" after Busy
- [+] (M) Dead Keys cause K95.EXE to terminate in NT
- [+] (K) If PDIAL is used, all subsequent calls are partially dialed
- [+] (K) HOST.KSC vs. K95 1.1.17
- [ ] (M) Don't tell K95 to RUN 16-bit programs
- [+] (K) SET SERVER IDLE-TIMEOUT doesn't timeout
- [+] (K) VT Origin mode does not apply to Next Line (ESC E) command
- [+] (K) PCTERM support not compiled into 1.1.17
- [*] (V) Using K95 with a NASI modem
- [*] (K) VT Soft-character-set assignments
- [ ] (M) Windows 95/98 DNS Caching results in Kerberos Authentication Failures
- [+] (C) TZ environment variable necessary for Kerberos Authentication
- [+] (K) Word Perfect Keyboard Mode does not work
- [ ] (M) More about Caps Lock
- [+] (K) CONNECT /TRIGGER causes memory exception in OS/2
- [ ] (C) "NO CARRIER", SET CARRIER-WATCH, INPUT and Hayes High-Speed Modems
- [+] (K) Install of K95 1.1.17 on Alpha does not perform registration
- [+] (K) Televideo terminals enter infinite loop when using Block mode
- [+] (K) SNI 97801 terminal emulation bugs
- [ ] (M) K95 Telnet can crash if remote host crashes
- [+] (K) K95 prints double characters when Copy Print mode is used
----(Version 1.1.18-CU)----
- [+] (K) REMOTE {SET, ASSIGN, RENAME, COPY} do not return to CONNECT mode
- [+] (K) Incoming TAPI calls may result in K95 crashes during file sends
- [+] (K) Kermit/2 1.1.17 server does not return from REMOTE HOST commands
- [+] (K) Transparent-print error if CSI received when remote character set is a code page
- [+] (V) MIT Kerberos 2.0 K5 support disables K95 telnet authentication
- [+] (V) SRP support fails to authenticate against SRP 1.5.0 Telnetd
- [*] (K) K95 1.1.17 breaks Host-Mode Script
- [*] (K) K95 and Host-Mode Script versions must agree
- [*] (K) TAKE command arguments are persistent
- [*] (K) Title-Bar Close box [x] doesn't work in Win9x
----(Version 1.1.19)----
- [+] (K) VT102 emulation broken in 1.1.19
- [*] (V) Failure to make certain Telnet connections
- [+] (K) Faulty OS/2 Dialer Patch for K95 1.1.19
- [+] (K) LOGIN.KSC didn't work without the Dialer
- [+] (K) C1 Printable characters of non-ISO-2022 character sets
- [+] (K) IBM 3151 attributes
- [+] (K) OpenSSL requires the PRNG to be seeded.
- [+] (V) Kerberos support not compatible with NRL SecureID support
- [+] (K) MAIL command does not work
- [+] (K) SEND ..\\*.c fails with "unreadable error"
- [+] (K) LOCAL and global variable name conflicts
- [+] (K) Kerberos password prompts versus -# 96 startup option
- [+] (K) ANSWER 0 improperly times out
- [+] (K) ANSWER fails to answer call with SET TAPI MODEM-DIALING ON
- [+] (K) RLOGIN Window Size reports are not sent
- [+] (K) Pulse dialing did not work
- [*] (V) AUTH K5 INIT /NO-ADDRESSES may crash Kermit 95
----(Version 1.1.20)----
- [+] (K) RLOGIN protocol (and secure variations) does not work in 1.1.20
- [+] (V) Kerberos 5 NT Alpha 2 and previous releases do not work with 1.1.20
- [+] (K) SET FILE DOWNLOAD-DIRECTORY leads to unexpected bahavior
- [+] (K) HOSTMODE scripts updated
- [+] (K) Telnet Forward-X option negotiated even when an X Server not found
- [+] (K) Wyse 60 keyboard definitions are incorrect
- [+] (K) \v(krb5errno) and \v(krb5errmsg) do not report all errors
- [+] (K) The Dialer does not support %ENVVAR% expansion of file names
- [*] (K) How to debug SET PRINTER configurations on Windows 95/98/NT/2000
- [+] (K) RLOGIN host userid fails to send userid to host
- [+] (K) -M command line option does not work
- [+] (K) \v(exedir), \v(startupdir), \v(inidir) may have spaces in the name
- [+] (K) Kermit 95 does not accept keyboard input on CJK Windows 95/98
- [+] (K) QNX terminal emulation keypad-plus and keypad-minus keys do not transmit "+" and "-"
- [+] (K) IBM 3151 unable to restore ASCII character set after Graphics character set is used
- [+] (K) TYPE and TRANSLATE commands only work under specific circumstances
- [+] (V) 1.1.20 is incompatible with Services for Unix 2.0 Telnet Server
- [*] (K) Wheel Mouse Button Numbering
- [+] (K) Transparent Print Bug introduced in 1.1.18
- [+] (K) Attempts to Add/Edit in K95 Dialer produce GPF on Windows 95/98
- [+] (K) K95 crashes on Windows 95 if SECUR32.DLL is missing
- [+] (K) K95 Dialer 1.1.20 dialer produces non-fatal GPF on startup
- [+] (K) Printer device not closed (if open) when K95 exits
- [*] (V) Winmodems
- [+] (V) SRP client security hole discovered
- [+] (K) K95 Documentation Shortcut to bugs file is incorrect
- [+] (K) UPDATES.HTM file has wrong version number
- [+] (K) K95 Linux Console emulation incompatible with Red Hat 7.x AND all other new Linux distributions
- [+] (K) Buffer overflow caused by long modem response messages
- [+] (K) Failures when connecting through SOCKS server takes too long
- [*] (V) Can't start K95 from a DOS Prompt or from with a DOS application
- [+] (K) K95.EXE when started from dialer complains about invalid command line parameters (Windows 2000/XP)
- [+] (K) K95D.EXE stops accepting new connections after several hundred or thousand incoming connections
- [+] (K) SET FILE COLLISION BACKUP fails on some file names but not others
- [+] (K) SET TAPI MODEM-DIALING ON can cause scripting to fail
- [+] (K) Dialer direct-serial configurations vs. Hardware Parity
- [+] (K) Windows XP default location causes K95.EXE to crash on startup
- [ ] (M) Ghost Character Fragments on NT/2000/XP Console Screen
- [*] (K) The Kermit Timesync Utility Might Not Work
----(Version 1.1.21)----
- [+] (K) Unbalanced-Braces Error in Scripts
- [ ] (K) Disappearing Cursor
- [ ] (K) Escape Character Status Line Typo
- [+] (K) Status Line Not Updated after Ctrl-] Ctrl-]
- [+] (K) HELP REMOTE HOST text wrong
- [+] (K) SET COMMAND SCROLLBACK in a script can cause an invalid memory access
- [+] (K) Kerberos 5 Rlogin fails to authenticate
- [+] (K) Dialer bugs
- [+] (K) telnet.exe and rlogin.exe do not work
- [+] (K) Telnet ComPort option
- [+] (K) Russian and Hebrew keyboard modes broken
- [+] (K) SHOW SSH does not list all forwarded ports
- [+] (K) COMPOSE and UCS2 Kverbs when used at Command Prompt do not
- [+] (K) SET SSH STRICT-HOST-CHECK ASK is treated as OFF
- [+] (K) \Kpaste on serial connections only sends 5 or fewer characters
- [+] (K) SET TERMINAL KEY { RUSSIAN, HEBREW } does not work
----(Version 2.0)----
- [ ] (?) Spurious K95G screen clearing on startup
- [ ] (V) Everson Mono Terminal Font Side Effects
- [*] ( ) Dimensions popup inactive during resize
- [*] ( ) Possible Help→Manual misbehavior
- [+] (K) Underline and Blink control
- [+] (K) Host-Initiated Resize vs Font Scaling
- [+] (K) SET GUI WINDOW RESIZE-MODE command missing
- [+] (K) Trouble with COM ports
- [ ] (K) Toolbar always visible
- [+] (K) Upgrade Registration Problem
----(Version 2.0.1)----
[ Top ] [ Index ] [ K95 Home ] [ Kermit Home ]
Ctrl-C typed in the Command screen to interrupt a command is not always
delivered to Kermit 95 by Windows, due to a bug in Windows 95 and Windows NT:
Microsoft PSS ID Numbers Q130717 and Q134284, 09-25-1995.
Once Ctrl-C fails to be delivered, it might never be delivered again.
Workaround: evidently the only workaround is to restart Windows. In the
meantime, you can still send Ctrl-C in the Kermit 95 Terminal screen using
Ctrl-Shift-C (or any other key to which you use SET KEY to map \3).
If you use Ctrl-C to interrupt a file transfer that has not yet started, there
may be a delay of several seconds before the Ctrl-C takes effect (this one is
fixed in 1.1.2).
In some cases, even when Ctrl-C is ignored, however, you might still be able
to use Ctrl-Break for the same purpose.
Also see: Bugs 83 and 152.
A workaround was discovered for the the Windows 95 bug that prevented K95 from
using screen widths other than 80 or heights other than 25, 43, or 50. As of
version 1.1.12, screens of any reasonable height and width can be used in
Windows 95 as well as Windows NT. OS/2 Warp 3.0 and earlier still restrict
console windows to 80 columns, but Warp 4.0 can use 132 columns.
Note, however, that Windows 95 video drivers are optimized for 80x(25,43,50);
other sizes work correctly as of K95 1.1.12, but might work slowly or exhibit
annoying amounts of flicker. This is a Windows 95 feature.
TAPI support was added in K95 1.1.12, and is documented in the current
edition of the K95 manual.
Kermit 95's registration procedure currently consists of customizing the
K95.EXE and K95DIAL.EXE programs with your name, company,
and serial number, and mailing back your registration card with one of your
Kermit 95 serial number stickers affixed to it. Future plans include online
registration.
In 1.1, registration was applied only to the K95.EXE file. Beginning
with 1.1.1 it was extended also to the Dialer.
NOTE: This item applies to Windows. In OS/2, loading code pages or
"PC fonts" into a full-screen session is not only permitted, but also safe --
provided the video driver allows it; evidently, however, the trend in new
video drivers is not to allow it -- thus preventing Kermit 95's Hebrew and
Cyrillic support from working as widely as it used to. See Section 3.4 of
OS2K95.TXT.
NOTE 2: As of version 1.1.8, this section applies only to Windows 9x.
The Windows NT version of Kermit 95 now uses Unicode fonts rather than
PC code pages.
As noted in the Kermit 95 manual, Windows 95/98 does not support code page
switching. The idea is that if you are going to use code pages, you should
use only one. But that's bad because some people really do need to work in
different languages; for example, English, Icelandic, Hebrew, and Russian.
Kermit 95 supports all of these very nicely as long as the underlying code
page is installed. Presumably if you are running Windows 95 in Iceland, your
code page is 850; in Poland it is 852; in Israel it is 862; in Russia it is
866; in Greece it is 869. So if you have the code page you need, no worries.
But if you need to work in (say) both Russian and Hebrew, you need a way to
switch code pages.
To change code pages, you can get a utility from Microsoft called CHANGECP.
It is on the Windows 95 CDROM, in the Other\Changecp folder. It
comes with lots of code pages, including 862 (Hebrew) and 866 (Russian). Its
menu only lists a few of them, but you can try typing any one you want on the
command line; for example:
changecp 850
or:
changecp 866
And then rebooting your system. You can't change code pages on the fly in
Windows 95 as you can in DOS. Nor is there any way to load fonts into a
DOS window of Windows 95 (such as the LOADFONT program we use with MS-DOS
Kermit to provide code-page switching).
Maybe CHANGECP will work for you, maybe not. When we tried it, 850 worked,
866 did not, even though all the supporting files for CP 866 were included.
The supporting files for Hebrew CP 862 are not even there, so this one didn't
even have a chance.
When you run CHANGECP, it prints a scary error message warning of possible
disk corruption. This means that if you use accented or special characters in
filenames (or for that matter, in the files themselves), the appearance of
those characters can change if you change code pages. If CHANGECP does not
work for you, our best advice is to give up -- do not try editing the Registry
or any other workaround, as this is almost certain to end in tears.
If a code page can be loaded, Kermit 95 can use it. But it seems that the
only way you can have a "non-Latin1" code page loaded is to install the
appropriate national version of Windows 95 -- such as Hebrew, Russian, Czech,
Ukrainian, Polish, etc.
If you want better language support in Windows 95 console applications,
express your wishes to Microsoft.
NOTE: You can use the DOS LOADFONT program together with the "PC fonts"
that are distributed with MS-DOS Kermit to load Cyrillic, Hebrew, and other
code pages into full-screen Windows sessions, BUT this tends to cause Windows
(not Kermit, but Windows) to crash, so this practice is not recommended,
encouraged, or supported. The true fix to this situation will come in the
conversion of K95 to full GUI, when it will no longer have to use code pages.
(As of version 1.1.8, the Windows NT console version uses a Unicode font, so
it can display any characters that are in its font -- e.g. Lucida Console.)
Certain keyboard scan codes, constructed by Kermit 95 from information
reported by Windows, are inconsistent between Windows NT and Windows 95.
These include Num Lock and Numeric Keypad /.
The Caps Lock bug is
fixed in 1.1.5. For dead keys, see items
103-104.
Also see items 283 and 386.
No problems in OS/2.
In the "Kermit 95" manual, Appendix I, Table I-5, pages 80-81, in the
"Assignment" column, all occurrences of "Ctrl-\" are incorrect and should be
replaced by "Ctrl-]", which is Kermit 95's default escape character.
Also, there are many new keyboard verbs, listed in KEYMAPS\KVERBS.TXT; the
default assignments are listed in KEYMAPS\DEFAULT.INI, and you can use the
new SAVE KEYMAP command to save your current configuration into a file.
As of version 1.1.14, the HELP KVERB command can be used to find out what
a particular kverb does and which key it is assigned to.
The Windows NT CMD.EXE shell bug that causes this problem is fixed in
Windows NT Workstation 3.51. This bug does not occur with replacement shells
such as JP Software's 4NT.EXE.
It's best to pick a specific size. When the setting is left at Auto,
sometimes the window will mysteriously and inexcplicably change size,
typically when returning to the Command screen from a non-standard-size
Terminal screen, often becoming a tiny small-print window. This also tends to
happen (but doesn't always) (in Windows 95 only) when you pick any size other
than 24 x 80 for a terminal screen on the Terminal page of the Dialer.
Starting with K95 1.1.12, the first version to support true 132-column screens
in Windows 95, a font size of Auto in the Console version can cause Windows 95
(original version) to create ever-smaller screens when switching back and forth
between 80- and 132-column mode, eventually resulting in alert boxes or
crashes. This is a Windows 95 bug, not a Kermit one. When this happens while
TAPI is in use, the modem will become unusable until Windows 95 is rebooted.
Choose a specific font size, not "Auto". See the instructions in the READ.ME
file creating a shortcut. You can choose your preferred font size in the
K95.PIF file that you create when following those instructions. (As noted
there, the default font size for console windows is set by editing the
Properties of C:\WINDOWS\SYSTEM\CONAGENT.EXE).
(Fixed.)
As of K95 1.1.19, no features are missing that are supposed to be present.
Of course many features remain to be added in future releases.
The behavior of Kermit 95 or any other program that uses serial ports will be
unpredictable if there are hardware conflicts. Typical symptoms include
application or system freeze-ups, general protection faults, etc. This is a
result of the PC architecture, which has a severely limited number of device
interrupts, but which is often overloaded with devices.
Serial communication devices must have unique addresses and IRQs (Interrupt
numbers). You can't have more than one device using the same IRQ. If you do,
neither device is likely to work -- or worse. Unfortunately, many devices
(such as network boards) are often shipped set to the same IRQ (3 or 4) that
is used by a serial port.
Serial ports COM3 and COM4 (or higher) are always problematic, since these are
not part of the standard PC architecture and there are no slots for them in
the PC BIOS. COM3 and COM4 are generally set by their manufacturers to use
IRQ 3 or 4, which conflict with COM1 or COM2. To resolve the conflict, you
must either find a free IRQ and set your COM3 or COM4 device to use it (and
tell Windows about it), or else disable one or more of the conflicting
devices. Changing the IRQ is risky because the new one you pick is likely to
conflict with something else, especially in a well-equipped PC (CDROM, sound
card, network adapter, etc).
In Windows 95, a COM port is supposed to be recognized automatically when you
install it. However, Windows might easily make incorrect assumptions about
its address or IRQ, or whether it has a FIFO (First-In-First-Out) buffer. You
can examine these, and correct them if necessary, in My Computer -> System ->
Device Manager -> Ports -> Advanced. In Windows NT, Control Panel -> Ports ->
Advanced.
Even when your configuration is conflict-free, you will experience data loss
-- often severe -- on serial connections if your serial-port controllers are
not "buffered": 16550A UARTs must be used, rather than lower-numbered models.
Unfortunately, many PCs are furnished with unbuffered UARTs by their
manufacturers as a cost-cutting measure. You can find out what kind of UARTs
you have on which COM ports, on which addresses, using which IRQs, in the
Windows dialogs mentioned above.
Steer clear of any modem that says "RPI" or "Controllerless" or "Winmodem" on
the box. Read the small print carefully. These are not fully functional
modems, and are very likely not to work adequately, or at all, under Windows
NT or OS/2. Although they are designed for Windows 95, they might not work
well there either, and even if they do, they might impose an unacceptable load
on your system, since the full job of the modem is being done by the PC CPU,
and/or they might interfere with your sound card. (Note: nowadays, most new
PCs come with Winmodems preinstalled -- the only way to avoid them is to
buy a second, external, modem and attach it to your COM1 or COM2 device.)
Another type of problem occurs whenever you use certain software that installs
its own drivers (such as COMM.DRV, the Windows 95 communications port driver)
over the real one. Such drivers are intended specifically for use with the
software they came with, and will often fail to work with other software.
Unfortunately, there is often no notification when a standard driver is
replaced in this manner, and also there is often no easy way to undo the
change. You can find out about your communications driver in the modem
diagnostics, Driver button. The communications port driver that comes with
the first release of Windows 95 is identified there as:
COMM.DRV, 5856 bytes, 7/11/1995 09:50AM
(the date and time display might vary with time zone).
Problems that are even more subtle and vexing can result from any of the
drivers you have installed on your PC -- SCSI-board drivers, sound-card
drivers, CD ROMs, etc. Sometimes the problems are solved by obtaining a newer
version of the driver from the board vendor. Other times, the only recourse
is to purchase an entirely different board and driver.
Even more ominous are problems with the PC BIOS. PC makers often have BIOS
upgrades available, perhaps downloadable from their BBS's or Web sites.
Also, owners of laptop and "green" PCs should watch out for the untoward
effects of power savers and suspend/resume features on internal modems,
PCMCIA cards, and similar hardware.
Finally, beware of any system changes that you make. In Windows, do not use
the Registry editor (REGEDIT) unless (a) you know what you are doing, and (b)
you have a way to back out of any changes. A seemingly innocent change in the
Registry can have unpredictable ramifications later on, causing Kermit 95 and
all your other applications, and Windows itself, to misbehave or malfunction.
In some cases the only recourse is to reformat your hard disk and reinstall
Windows 95 from the original media.
- The Hangup button does not interrupt dialing. If you need to interrupt
dialing, bring the Kermit 95 Command window to the front and type Ctrl-C.
- Other items from this entry are
fixed.
- The Terminal screen can become frozen under certain circumstances; for
example, receipt of an unterminated escape sequence (perhaps caused by
noise). When this happens, the Terminal screen will report OSC, APC,
DCS, etc, in the status line. To unfreeze, use Alt-R (Reset) while in
the terminal screen. This is not a bug -- it is how real VT 220 and
higher terminals work.
- If you try to download a file directly to the printer by naming the file
"PRN", this will fail unless you "set file collision overwrite". That's
because Kermit's normal action when receiving a file is create a backup
copy -- but it can't make a backup copy of the printer :-) Similarly for
other files whose names are the same as those of DOS devices: LPT1, AUX,
COM2, and so on.
- The printer driver might fail to print the last or only page. This is a
traditional shortcoming of PC operating systems. Sometimes the page can
be forced out by using Alt-E (verb \KprintFF) to send a formfeed to the
printer, or by sending another print job, to the printer. Some printers
(such as the HP4L and HP5L) have a special button that ejects the page.
K95 1.1.14 and later offer additional printer controls, such as
the /END-OF-JOB switch, documented in the current K95 manual.
You can save the configuration for a particular Dialer entry into a plain-text
file by highlighting the entry, then going to File on the main menu and
choosing Generate Script File. The script file goes into Kermit 95's SCRIPTS
subdirectory and has the same name as the highlighted Dialer entry, but with
spaces and special characters converted to underscores. It is a plain-text
file composed of Kermit commands that can be easily faxed or e-mailed to our
tech-support staff.
You can troubleshoot connection problems by following the instructions in
Chapter 3-6 of Using C-Kermit, second
edition. You can obtain transcripts and logs of troublesome connections using
"log session" and also with "set dial display on" and then using your mouse to
copy and paste the modem dialog from the screen into a file.
You can troubleshoot terminal emulation problems by putting the terminal
screen into debug mode (with Alt-D when the terminal screen is active).
You can troubleshoot file-transfer problems by following the instructions in
Chapter 10 of Using C-Kermit (second edition). File-transfer
packet logs are obtained with the "log packets" command.
The ultimate debugging tool is the debug log, obtained with "log debug".
It tends to grow huge rather quickly.
Any or all of these logs can be sent to our tech support staff for analysis
in cases where suggested remedies or workarounds are ineffective.
Version 1.1.14 and later allow the session.log and debug.log files to be
activated in the Dialer entry.
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.) However, as of K95 1.1.19,
the BBS is being retired due to lack of use.
Kermit 95 is necessarily less snappy than MS-DOS Kermit (under DOS) on the
same type of PC because MS-DOS Kermit is in direct control of the
communications and the video, whereas Kermit 95 must go through several
layers of drivers and operating-system calls to do the same thing.
Video drivers vary in speed. Those in Windows NT tend to be significantly
slower than those in Windows 95. Even in Windows 95, some drivers or
hardware might be slower than Kermit 95 was designed for. Kermit 95's
default screen-updating cycle is 100 milliseconds (1/10 second); if your
video driver is slower than that, you might be able to make Kermit go faster
by increasing (yes, increasing) its screen-update interval, for example:
SET TERMINAL SCREEN-UPDATE FAST 200
Flickering can occur when the video driver is not performing a comparison
between what is already on the screen and what it is being asked to write.
So it paints the entire window (which takes time) even when it doesn't have
to. When acceleration is on (and supported by the driver) minimal screen
paints occur and screen updates should be fast. In the Control Panel either
on the Display object or the System Object, there is a performance page which
allows you to set the Video Acceleration properties. Make sure this is set
to use full acceleration. The only flickering you should see in this case is
the mouse pointer, since it is turned off and on before and after each
console screen paint.
If that doesn't help, you can experiment with Kermit 95's SET TERMINAL
SCREEN-UPDATE FAST and SMOOTH options to choose which algorithm is used to
update the screen when new data arrives from the host. Whenever a byte
arrives from the host it is processed by an incoming data thread. If it would
result in a screen write, an internal screen buffer is updated and a flag is
set. When using the FAST algorithm, the screen buffer is copied to the
physical screen (if the flag is set) on a timer thereby allowing the maximum
data throughput from the host to occur regardless of the speed of the video
drivers. This is the default screen updating method.
The negative aspect of this method is that on really fast connections it is
possible for some screen writes to never make it to the screen, e.g. during
continuous scrolling (but the data is processed correctly, and can be viewed
in the scrollback buffer). The SMOOTH option says to copy the screen buffer
to the window whenever the dirty flag is set. This results in much slower
data handling, and a potentially significant increase in the number of screen
writes, but on the other hand it tends to make echoing snappier when you are
typing. Also see item 469.
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
In Windows 95, but not Windows NT or OS/2, if a REMOTE HOST command is sent to
a Kermit 95 server, the server will lock up. Diagnosis: The server end of
REMOTE HOST commands is implemented by running the corresponding operating
system command with its standard output piped back to the server; redirection
of standard output in Windows 95 can cause the parent process to get stuck in
a read that never returns (refer to Microsoft Knowledge Base Article Q150956).
Partially fixed in version 1.1.3, by using temporary files rather than pipes.
Partially fixed even more in 1.1.5 by recoding REMOTE DIRECTORY, REMOTE
DELETE, etc, not to call the Windows shell, but to do everything internally.
So now, hopefully, the problem occurs only with REMOTE HOST commands, if at
all.
Fixed even more in 1.1.16 by changing the handle used by K95 for accessing
the console device.
(Also see item 177.)
Kermit initialization and key mapping files historically have a filetype
(extension) of .INI, but these are different from Windows .INI files.
Kermit online plain-text documentation have historically been had the .DOC
filetype, but Windows thinks these are Word documents. In Kermit 95 1.1.11,
all *.DOC files were renamed to *.TXT.
Similarly, Kermit script program files have long had a filetype of .SCR, but
in Windows 95, .SCR files are associated with screen savers. Thus if you
double click on a Kermit .SCR file, you are likely to get an undesired
result.
In Kermit 95 1.1.3, all script files distributed with Kermit 95 were renamed
from *.SCR to *.KSC, and all references to them (e.g. in the many Dialer
pages) were changed accordingly, and the new script files that were included
with this release for host mode were also given the .KSC extension. The .INI
files, however, were not renamed. Kermit has been around a lot longer than
Windows!
(Fixed.)
The K-95 icon is in the Windows 95 icon format, which can't be correctly
displayed in Windows NT 3.xx; it shows up as a black square. It works fine
in Windows NT 4.0. (An OS/2-format icon is included with the OS/2 version.)
(Fixed.)
Certain files mentioned in Kermit 95 manual are missing. These include:
- P.DOC (after the manual was written, the P.EXE program
for XYZMODEM protocol file transfers was fully integrated into Kermit 95, and
thus has no user interface or commands of its own);
- KERMIT.KDD (the Kermit Dialing Directory) referred to on page 30
is really CKERMIT.KDD in the PHONES subdirectory;
- DOCS\FONTS.DOC does not exist. At press time we thought we had a
safe way to switch fonts, but it was not to be. A forthcoming release will
address this issue.
- DOCS\APC.DOC does not exist. The APC feature is fully documented
in the second edition of Using C-Kermit.
(Fixed.)
Restriction. The Winsock connect() function does not respond to the
WSACancel call.
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
"If I put text into the Windows Clipboard using regular Windows services,
such as the Mark and Copy buttons on a console-window Toolbar, or
mouse-select and Edit->Copy in a GUI Window, then I can't use the Paste
button on the Kermit 95 window toolbar to paste the material into the Kermit
95 screen." Cure: Open the Properties dialog for the window; go to the Misc
page and turn off "Fast Pasting".
No problem in Windows NT. In OS/2, Kermit 95 comes with K2CLIP.EXE, the
Kermit/2 Clipboard Server, to take care of the clipboard.
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
The Microsoft VCOMM driver used in Windows 95 versions prior to OEM SR2 for
serial port services in console windows does not support Plug-n-Play.
Therefore, 32-bit communication programs such as Kermit 95 can not use these
modems in Windows 95 unless they are previously initialized by another
program, such as Hyperterminal or Dial Up Networking.
Version 1.1.10 of Kermit comes with a new utility, K95CINIT, to initialize
such modems so they can be used by Kermit 95. This program needs to be run
only once each time you boot Windows 95. You can run it "by hand" at any time
before starting Kermit 95, or you can put put in your StartUp Menu to make it
run automatically each time you restart Windows.
Version 1.1.12 supports TAPI and so should work with any modem or other TAPI
device at all; K95CINIT is not needed in K95 1.1.12 or later unless you want
to use a Plug-n-Play COM-port device directly that Kermit can not normally see
on its own as a COM port, rather than through TAPI.
K95's RLOGIN connections seem not to work at all with FTP Software OnNet-32.
To enable them, you have to edit the ONNET32\ETC\SERVICES file and comment out
the definitions of "login" service on port 49. The port that works is 513.
Now you can make RLOGIN connections, but certain RLOGIN features, such as
automatic screen-resize notification, do not work because OnNet-32 fails to
properly deliver RLOGIN protocol messages as out-of-band data to K95. The
same code works fine with Microsoft and OS/2 TCP/IP.
You can also specify the port number (not name) in RLOGIN commands to Kermit:
rlogin hostname:513 [ userid ]
This allows you to make RLOGIN connections without editing the SERVICES file.
An IP address of 0 can sometimes be used to indicate one's own IP host.
Using this construction in K95 with OnNet-32, however, causes K95 to crash.
Diagnosis: Bug in OnNet-32. Workaround: Don't "set host 0" or "telnet 0"
in OnNet-32.
Digital UNIX 3.2 (OSF/1 3.2) seems to think your terminal has one more line
than it really has and so "man", "more", and similar commands try to use
Kermit's status line for screen data. Workarounds -- either:
- Tell UNIX "stty rows 23" (one less than Kermit's screen length), or:
- Tell Kermit "set term height 25".
- Tell Kermit "set term statusline off" (1.1.5) to let UNIX use the
bottom line.
It is not known whether Digital UNIX 4.0 behaves differently.
On certain PCs, when the system load is high and/or when memory, swapping, or
other resources are low, Windows's screen painting algorithm sometimes misses
the fact that some areas of the screen need to be updated. Diagnosis: Bug in
Windows 95 CONAGENT.EXE, which tries to optimize screen painting by comparing
old and new window contents. Sometimes it misses. Workaround: Use Alt-Enter
to switch between window and full screen will temporarily clear up the
problem. Also, rolling back and then forwards will also tend to force the
screen updates to be applied correctly. Workaround: adjust the video
optimization using Control Panel -> System -> Performance, or adjust the
resolution or color depth to reduce CPU requirements. Cure: Must be fixed by
Microsoft. Also see Bug 238.
(Fixed.)
(Fixed.)
If the "enter dial mode" string causes an error, use the new command
"set modem command dial-mode" to alter or remove it.
(See bug 47.)
Uploading files on a LAT connection can be problematic on some connections,
but not others. If you get a lot of retries or failures, try using 90-byte
packets and 1 window slot (tell VMS C-Kermit to "set receive packet-length
90"). Downloads can use any packet length or window size. WARNING: Do NOT
tell VMS or VMS C-Kermit to disable flow control. VMS C-Kermit MUST have
"set flow xon/xoff". On other connections, uploads using long packets and
sliding windows work just fine. No pattern has yet been discovered to
explain success or failure.
(Fixed.)
Kermit 95 1.1.12 offers various new SET PRINTER options
that should address this problem. See the current manual, or type HELP
SET PRINTER at the K95 prompt, for details.
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
One file name in the Kermit 95 distribution, K95CUSTOM.INI, exceeds the 8.3
FAT name format. This is not a problem in Windows 95, but it is bad news in
Windows NT when long filenames are disabled (by editing the registry) -- the
installation procedure fails. Even if the industrious user finds a way to
complete the installation, the temporary files created by the Dialer (based on
the Dialer entry names, many of which are quite long) will cause runtime
failures. So, at least for the present, maybe forever, this is a restriction.
Kermit 95 can not be installed on file systems restricted to 8.3 notation.
VT320 and Wyse host-controlled status line features are not fully implemented.
This will come with the GUI release. However, in version 1.1.5 it is possible
to remove Kermit's status line from the screen, making it available to the
host; this satisfies the requirements of many host-resident programs for an
extra status line. (The command is SET TERMINAL STATUSLINE OFF.)
Version 1.1.8 adds full support for the VT320 status line (and for user
messages on the Wyse models and Televideo label lines).
This capability was added in version 1.1.8, but only if you enable it first
with SET TERMINAL SEND-DATA ON. Use at your own risk -- host-initiated
screen transmissions are a notorious security loophole.
(This entry applies only to Windows 95)
The bugs noted in Entry 6 regarding dead keys and Caps Lock should be mostly
fixed in version 1.1.5 by doing an end-run around the buggy Windows 95 console
agent(*), and adding our own substitute keyboard handling in its place. The
fix is not perfect, however, because the "raw" keyboard driver is still hidden
from us, and the information we do receive is partially corrupted by Windows.
On the positive side, the Caps Lock bug is gone, and dead keys seem to work
correctly on US systems when the "international" keyboard driver is selected,
which uses dead keys, as well as on real German, Finnish, and other West
European Windows 95 systems, where it is also reported that the special
single-stroke keys (such as Umlauts on the German keyboard) work correctly.
The the terminal emulation and character-sets sections of the current K95
manual for details.
On the negative side, the fix interferes with the Alt-Gr key which, on most
non-USA keyboards, is used to select the "third" (or lower right, or front)
character printed on the keytop. Alt-Gr is normally the right Alt key, and it
is really just a shorthand for Ctrl-Alt. Thus, for example, on the German
keyboard, Alt-Gr-q produces at-sign (@), and so does Ctrl-Alt-q. To work
around this side effect, you can either use Ctrl-Alt instead of Alt-Gr, or
you can give this command to Kermit 95:
SET WIN95 ALT-GR ON
This tells Kermit 95 to treat the right Alt key as if it were Ctrl-Alt.
If SET WIN95 ALT-GR ON produces other unwanted and unanticipated side effects,
set it back OFF and use Ctrl-Alt instead.
SET WIN95 ALT-GR ON enables the use of Alt-Gr as a SHIFT key (i.e. to be held
down at the same time as the key it is modifying, just like Shift, Alt, or
Ctrl). Kermit 95 does not support the use of Alt-Gr as a "prefix" or "sticky
shift" key (to be pressed and released prior to pressing the modified key).
Also, our workaround has moved some Control keys. For example, Ctrl-\
(ASCII 28, Field Separator) which is normally sent by Ctrl-sz (Ess-Zet) on a
German keyboard (top left of 0 in digits rank), or Ctrl-] (ASCII 29), normally
sent by Ctrl-9, are now sent from their original positions on the US keyboard.
If you want to change these back to their expected positions, use SET KEY.
Of course, under Windows NT, dead keys, special keys, and Alt-Gr work
naturally and correctly (without any workarounds in the Kermit code), except
that some NT users report a beep whenever a dead key is pressed (even though
the correct dead-key combinations are still produced).
All of these problems will disappear when the GUI version is released and
Kermit is no longer accessing any of the bug-ridden Windows console services.
(*) Microsoft PSS ID Q140456, 12-07-1995: "Deadkeys cannot be accepted as
input to a Windows 95 Console application even when the appropriate keyboard
layout is selected. ReadConsoleInput() cannot read deadkeys in
Windows 95".
NOTE: Alt-Gr and Dead-key handling were broken again during the translation to
Unicode in 1.1.8 and was fixed and improved in 1.1.14. As of 1.1.14, K95
should properly default the SET WIN95 ALT-GR command for most keyboard
layouts. 8-bit characters should now be properly translated for all locales
from the ANSI code page to the OEM code page used by the Windows Console
environment.
NOTE: This item applies only to Windows 9x. Since the release of version
1.1.7 we have had confirmation from Poland and other East European
countries that the features described below work as advertised.
Testing of version 1.1.5 on East European Windows 95 versions was limited to
the Czech version. Results are unknown for Polish, Hungarian, Romanian,
and other East European versions. If you have problems entering East-European
(Latin2) characters, first make sure you have:
set terminal character-set latin2 cp852
set terminal byte 8
If SHOW terminal doesn't say your code page 852, then also:
set win95 keyboard latin2
and, as noted in item 103:
set win95 alt-gr on
In the Terminal Window, you can also enter East-European characters using
Kermit 95's Latin-2 Compose-key feature, documented in the current manual.
As of 1.1.14, SET WIN95 KEYBOARD-TRANSLATION is no longer needed. Translation
from ANSI code pages to OEM code pages is performed using Win32 API calls.
(Fixed.)
(Fixed.)
We never said they did. Not only that, K95's terminal-to-host reports
notify the host that they are not supported. This feature can not be
implemented in a Console program. It is on the list for consideration in the
full GUI version.
(Fixed.)
(Fixed.)
(Fixed.)
There are all sorts of bugs in Windows 95 regarding support of console
applications (like Kermit 95), particularly when they are started from GUI
applications (like Netscape). One of many typical symptoms is loss of
keyboard focus. One simple workaround (not guaranteed, but seems to work in
many cases) is to have Netscape start the Kermit 95 TELNET.EXE, which
in turn starts K95.EXE; thus K95.EXE is not being started by
a GUI application, and therefore this bug MIGHT not be tickled. If that
doesn't work, try making a PIF file for K95.EXE and then having
Netscape run the PIF file, rather than the EXE.
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
NT bug. Workaround: If this happens to you, try using Ctrl-Break instead of
Ctrl-C. (The problem should now be fixed by adding a separate Ctrl-C handler
to each thread.)
(Fixed.)
Kermit 95 for OS/2 includes a SOCKS4 interface, described in
the Kermit 95 for OS/2 notes.
The Windows version does not. Firewall support needs to be built into the
TCP/IP stack, or the network itself (e.g. in a router or transparent bridge),
not each application. Microsoft Winsock, for example, does not support the
SOCKS firewall negotiation mechanism, but Trumpet Winsock and FTP OnNet-32 do
support it.
The Microsoft Catapult proxy server announced in July 1996, however,
is supposed to provide transparent firewall support, both CERN and RWS (Remote
WinSock) styles, to all Windows applications, and so should work with K95.
And from NEC, we have a new 32-bit version of SocksCap, which sits between
Windows Sockets (WinSock) and the application (Kermit 95, Netscape, etc). Its
authors have tested it successfully with 32-bit Microsoft WinSock on both
Windows 95 and NT using many different applications. It should work with
SOCKS4 and SOCKS5 servers. You can find SocksCap at:
http://www.socks.nec.com/sockscap.html
ftp://ftp.nec.com/pub/socks/sockscap/
Security concerns are addressed in K95 1.1.16 for Windows 95 and NT by
addition of Kerberos IV and V authentication and encryption.
CLICK HERE for further information. Kerberos
authentication as well as CAST and DES encryption carry export restrictions
mandated by USA law.
(Fixed.)
The TCP/IP stack in LAN WorkPlace 5.0 is 16-bit. Kermit 95 only works with
32-bit Winsock stacks. Workaround: You can use MS-DOS Kermit with the LWP
16-bit stack (via its TELAPI interface); or obtain a 32-bit stack from Novell,
if they have one.
Serial speeds such as 75/1200 can't be used in Windows. They are not
supported by the serial driver.
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
Holding down the Alt key and pressing 1 to 4 consecutive digit keys on the
numeric keypad, while Num Lock is on, and then letting go of the Alt key, is
supposed to return the character whose code value was typed. This is supposed
to be a way of typing special characters that can't be typed any other way on
your keyboard. Kermit 95, however, sees each key as you press it. Thus,
Alt-<123> returns three scan codes -- Alt-Keypad-End, Alt-Keypad-Downarrow,
and Alt-Keypad-PgDn -- rather than the single character whose code value is
123. If you need to type a character that is not available on any key, use
SET KEY to define a key to the desired code, e.g. when using Code page 437 to
assign lowercase letter i with grave accent to Ctrl-Shift-Equals:
set key \1979 \161
This works in the Terminal window and, as of version 1.1.5, also in the
command window. As of K95 1.1.16, you can also use SET TERMINAL KEY /LITERAL
to enter a literal definition, i.e. one that will not be translated.
Interruption of X-, Y-, and ZMODEM transfers by typing X, Z, E, etc, when the
file transfer screen is active, is not implemented yet. Interrupting with
Ctrl-C works, but it leaves the other XYZMODEM program in protocol mode; to
take it out of protocol mode, you have to get back in the terminal window and
type five (5) Ctrl-X characters in a row. In version 1.1.6, the other ZMODEM
program should exit automatically when you Ctrl-C Kermit 95, so the five
Ctrl-X's should not be needed.
Some modems have a feature called adaptive dialing. When they are told to
dial a number using Tone dialing, they check to make sure that dialtone has
gone away after dialing the first digit. If it has not, the modem assumes the
phone line does not accept Tone dialing and so switches to Pulse.
When dialing out from a PBX, there is almost always a secondary dialtone.
Typically you take the phone off-hook, get the PBX dialtone, dial "9" to get
an outside line, and then get the phone company's dialtone. In a situation
like this, you need to tell the modem to expect the secondary dialtone. On
Hayes and compatible modems, this is done by putting a "W" in the dial string
at the appropriate place. For example, to dial 9 for an outside line, and
then 7654321, use ATDT9W7654321. In Kermit 95, this is accomplished with:
SET DIAL PBX-OUTSIDE-PREFIX 9W
(replace "9" with whatever your PBX's outside-line prefix is).
(Fixed.)
(Fixed.)
In Windows 95 and NT, the sounds that K95 uses are as follows:
- Terminal Bell
- Beeps the speaker (this is sometimes mapped by the
sound drivers to the Windows Default Sound as
configured in the Control Panel.
- Beep Information
- Windows Default Sound
- Beep Warning
- Windows Asterisk
- Beep Error
- Windows Exclamation
If these sound events do not have .WAV files associated with them in
the control panel they will not generate any sound.
Also, make sure that you have speakers hooked up to your computer.
Many new computers with built-in stereo sound boards do not have
internal speakers.
(Fixed.)
(Fixed.)
Russian, Hebrew, and EMACS mode layouts could not be modified in 1.1.7 and
earlier. This is fixed by the addition of the SET TERMINAL KEY command in
version 1.1.8.
As noted in Bug 103, the fixes that were added to work
around the problems with Caps Lock and dead keys in the Microsoft console
agent might cause some of the control keys to "move" on national keyboards,
such as the German ones, from the position indicated by the keytop label to
the US-keyboard position. Reportedly, this is nothing unique to Kermit; it
also happens with other communications software too, including Microsoft's own
Telnet and Hyperterminal. Here is a sample key mapping file to put the
control keys back where they belong on a German keyboard, courtesy of Josef
Hinterreger:
; GERMAN keyboard
set win95 alt-gr on
; control key definitions
; in addition to predefined US-positions
set key \1336 \27 ; Ctrl-[ (Escape)
set key \1499 \28 ; Ctrl-\
set key \1337 \29 ; Ctrl-]
set key \1054 \30 ; Ctrl-^
set key \1334 \30 ; Ctrl-^
set key \1500 \30 ; Ctrl-^
set key \1469 \31 ; Ctrl-_
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
Commands can not be edited with arrow keys, Insert/Delete, Home, End, or other
PC keys, or the mouse. Nor can you use EMACS or VI style keystrokes to edit
commands. Command editing is accomplished as described in "Using C-Kermit",
with the addition of up- and down-arrow keys (or Ctrl-B and Ctrl-N) for
command recall. Restriction.
The DIRECTORY and TYPE commands are built in to get around Windows 95 bugs
with pipes and process control. If you would rather have Windows or OS/2
produce directory listings or type your files, then use "run dir" or "run
type" or "run more", etc. In that case, you can also include Windows or OS/2
shell command switches. But watch out for loss of keyboard focus and other
problems that can occur when you run inferior processes this way, especially
if you interrupt them with Ctrl-C or Ctrl-Break.
(Fixed.)
(Fixed.)
(Fixed.)
Several users have reported that the command screen loses messages or becomes
scrambled after giving a DIAL command, or two DIAL commands, or an ANSWER
command followed by a DIAL command, or after interrupting a DIAL command with
Ctrl-C. Similarly, if you interrupt a file transfer with Ctrl-C and then
start another one, the screen is not fully repainted. This problem started
happening in version 1.1.5 when we switched the command screen to use the
same virtual screen management that used by the terminal screen, to allow
scrollback, etc. The cause is probably related to
Bug 83 above, which is
evidently exacerbated when an asynchronous process-related interrupt (like
Ctrl-C) occurs. If this happens to you in the file-transfer display screen,
use Ctrl-L to refresh (or at least partially refresh) the screen.
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
Evidently version 1.1.3 was distributed in more than one form, so patching
over the wrong form won't work. Sorry, our mistake -- if this happens to you,
contact us.
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
The patch product used for applying patches to Kermit 95 only understands DOS
(FAT, 8.3) filenames on its command line. You can't give a filename to the
Patch program that does not conform to FAT restrictions: directory names must
not be longer than 8 characters, and must not contain spaces; filenames must
have no more than 8 characters before the dot, and no more than three after
it, and must not contain spaces, and there can not be more than one dot.
Examples of patch commands that work:
patch wi115-6
patch d:\k95\download\wi115-6
Example of a patch command that doesn't work:
patch D:\Downloads\Kermit 95 Patch\K95-Intel-from-1.1.5-to-1.1.6
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
In output-file specifications only, it is ambiguous whether
\digit (like \4DOS) combinations are literal or are
the backslash-encoding of special characters. So, for example, SET
PRINTER DOWNLOAD\123.TXT does not work as expected. If you need to refer
to a directory name that starts with a digit, a percent sign, an ampersand, or
other character that signals a C-Kermit backslash code, in an output-file
specification, either use a forward slash or double the backslash, e.g.:
SET PRINTER DOWNLOAD/123.TXT or SET PRINTER
DOWNLOAD\\123.TXT.
(Fixed.)
(Fixed.)
Several Kermit 95 users have reported that the K95 Dialer suddenly and
mysteriously stopped working. Diagnosis:
- Kermit 95 Dialer database files (DIALER.DAT, DIALUSR.DAT, DIALINF.DAT) have
an extension or filetype (the part of the filename after the period) of DAT
(not unreasonable; people have been using DAT for data files for decades).
- Certain other applications, when you install them, register themselves in
the Windows Registry as the application for all files having the .DAT
extension. These applications include at least ACT! (from Symantec) and
Wordperfect 7 (from Corel).
- The Kermit 95 Dialer database files now appear with the icon of the other
application. Should you click on one of these icons, the other application
will be started and fed the K95 database file, and will likely damage it.
If that happens, the Dialer might be unusable from that point on.
There is nothing Kermit can do to prevent this. There is no mechanism for
arbitration of conflicting file associations. If filetype XYZ is associated
with application A and then you install a new application, B, it might undo
application A's association and take over its files (or, conversely, it might
find that its files belong application A). This is intrinsic to the design of
Windows 95 and NT. Only one application can be the default application for
each filetype.
To defend against this effect, make a backup copy of your *.DAT files prior to
installing any new application, preferably on a diskette, which you should
remove immediately and put in a safe place. And then remember to avoid
clicking on any of the files in your Kermit 95 directory, except of course the
EXE files, no matter what interesting icons they might display. (This warning
really applies to ALL your applications, not just to Kermit 95 -- never click
on a data file's icon if the icon does not seem appropriate.)
As of version 1.1.7, the Dialer keeps two backup copies of each .DAT file
as changes are made.
(Fixed.)
(Fixed.)
This is not really a bug; we never said they were, but we never said they
weren't either. Status: To be added in a future release, maybe. In fact, all
of these are supported internally, but there is no command to invoke them; the
XYZMODEM program on the other end can request any of these and it will be used.
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
Beginning in version 1.1.5, K95.EXE's standard input can not be
redirected. This works around a bug in Windows 95 regarding invocation of
console programs from GUI programs (e.g. K95.EXE from Netscape).
BUT... Beginning in K95 1.1.16, redirection to/from a pipe is automatically
detected, and there is a new command-line option to force the use of stdin
and/or stdout, documented in the current K95 manual.
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
See the original description of this bug.
In Kermit 95 1.1.19, we work around it by disabling the Close box.
Startup of the K95 Dialer is unavoidably slow because it makes backup copies
of all of its database files every time it starts. The backup copies go into
your PC's temporary directories (identified by the TEMP or TMP environment
variable). The speed of startup depends on the speed of the disk that is used
and the sizes of the database files. If you don't need all the predefined
BBS and Telnet site entries, etc, see the READ.ME file for instructions on
removing them.
In version 1.1.8, an 8th button -- "ScriptFile" -- was added to the Toolbar.
This button might wrap around to a second Toolbar line if you are installing
1.1.8 as a patch to a previous version. If this happens to you, you can widen
the main Dialer window by stretching it with the mouse. In version 1.1.14
the ScriptFile button was renamed to ShortCut because now it creates a
clickable shortcut on your desktop and/or in your Start menu.
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Windows only) "Every time I start K95 to make a Telnet connection, the
dialup networking program is launched automatically. I have to cancel it
first and then continue logging in." This is not a Kermit problem -- the
same thing would happen with any Internet application, such as Microsoft
Telnet or FTP. To fix: open your Windows 95 Dial Up Networking folder
(should be in your "My Computer" folder), go to Connections->Settings->
General and choose "Don't prompt to use Dial-Up Networking".
If that doesn't work and you are using Windows 95b (OSR2), check the setting
for "auto connect" in the Internet Control Panel.
"Why does the status line say Script instead of Prompt sometimes?" It says
this when CONNECT mode was entered from a script (command file or macro)
rather than from the prompt; thus escaping back from CONNECT mode will
continue the script (if there are any commands left in it), rather than
returning directly to the prompt.
There are two likely explanations for missing login prompts:
First, it can happen because Kermit 95 sent your user ID in advance. This
happens on Rlogin connections and (in 1.1.8 and later) also on TELNET
connections if the Telnet sever supports the NEW-ENVIRONMENT protocol and
Kermit 95's LOGIN USER variable is set to match your username on the host.
Just enter your password, or defeat this feature with SET LOGIN USER to
nothing.
Second, this can happen when TELNET terminal-type negotiations cause K95 to
change its terminal type (and therefore to clear the screen); use K95's SET
TELNET TERMINAL-TYPE command to supply a terminal name acceptable to the host
and to suppress automatic terminal-type switching.
In version 1.1.16, this also happens when you make a connection authenticated
by Kerberos or SRP, in which case even the Password: prompt would be missing.
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
Windows 95 QuickEdit (selected in the console window Properties notebook,
General page, under Mouse), gets in the way of Kermit 95's mouse functions.
Restriction: You can have one of these or the other, but not both.
If you use the Caps Lock key to change the Caps Lock state from what it was
when K95 was started, and then select another window, the Caps Lock state goes
back to what it was when you entered the K95 window. Diagnosis: bug in
Microsoft CONAGENT.EXE. Happens only in Windows 95, not in Windows
NT.
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
The block character used on the left and right end of the status line to
indicate the end of the left-right scrolling region in 1.1.9 and later comes
out too small in certain fonts or sizes but not in others. When it's too
small, it looks like a fifth LED rather than a distinct symbol. For example,
it's a nice large block in TrueType 10x20 or 8x16 but it's too small in 10x18
or 8x8.
You can override the
choice of editor by adding the appropriate SET EDITOR command to your
K95CUSTOM.INI file, or by having an EDITOR environment variable defined at
the time K95 starts up. Remember, the SET EDITOR command (and the EDITOR
environment variable) must contain the full pathname of the editor program,
such as C:\WINDOWS\COMMAND\EDIT.COM. Also, the "netedit" process, especially
when host-initiated, results in a lot of screen flashing as K95 switches back
and forth between its command screen, file-transfer display screen, and
terminal screen. It can't be avoided. Also see item 268.
NOTE: In version 1.1.12 and later, the full pathname of the editor is not
required if the editor program is in the PATH.
(Fixed.)
(Fixed.)
In an interrupt driven system such as Windows 95, Windows NT, UNIX, etc, the
keyboard driver is called to handle each event as it occurs. When the system
is under heavy load, the events cannot be processed in real time and are
placed on an event queue. In Windows 95, CONAGENT.EXE has a bug that
causes the event queue to be processed in last-in-first-out rather than the
expected first-in-first-out order. Thus fast typists might occasionally see
their keystrokes transmitted in reverse order. And (you guessed it) this does
not happen in Windows NT or OS/2.
If you have installed Compuserve, America OnLine, or any other browser
package from an Internet Service Provider that uses its own proprietary
Sockets Library, there is a good chance that your existing Winsock
applications (including, but not limited to, Kermit 95) will no longer work.
This happens when the installation either replaces the Windows version of
WINSOCK.DLL and/or WSOCK32.DLL or adds new versions in a new directory.
While all Winsocks are supposed to implement the same API, they don't always
implement the same set of features or have the same bugs. Therefore,
applications like K95 that make rather full use of Winsock might not work
properly with these replacement DLLs.
These applications and service providers should not be replacing the standard
Windows TCP/IP drivers. This is an extremely reckless act and all bets are
off once they have done it. If you are affected by an application that has
changed Windows' TCP/IP stack, contact the vendor of that application for a
way to safely back out of the installation.
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
Sorry, there's not enough room on the screen. To be addressed in GUI release.
If you have an environment variable SHELL or COMSPEC, and it is NOT defined
to be the path of COMMAND.COM or other valid command shell, then PUSH, RUN,
and related commands will not work.
(Fixed.)
(Fixed.)
There is a type of terminal emulation commonly called ANSI, which is used by
BBS's. It is very simple -- essentially nothing more than what you get with
ANSI.SYS on a DOS PC. Then there is the highly evolved and complex version of
it used on the SCO (and Unixware) console. Kermit 95 supports both; the first
emulation is called ANSI; the second is called SCOANSI. However, the *name*
SCOANSI is unknown on SCO systems (SCO Xenix, SCO UNIX, ODT, and Openserver
versions 5.0.4 and earlier, and Unixware versions prior to 7.0), where it is
called simply ANSI. So if a Telnet client such as Kermit 95 comes in
announcing its terminal type as SCOANSI, the SCO host doesn't recognize it.
In Kermit 95 1.1.8 and later, the Telnet terminal-type negotiation feature
will cause Kermit 95 to change to another terminal type, most likely "plain
old" ANSI. To work around this problem in Kermit 95:
set terminal type scoansi
set telnet terminal-type ansi
set terminal character-set transparent
Or on the SCO system (Reference: SCO Technical Article 109521, 17 July 1998):
# vi /etc/termcap (on the pre - 5.0.5 system)
Find the line that reads:
li|ansi|ansi80x25|Ansi standard crt:\
and change it to read:
li|ansi|scoansi|ansi80x25|Ansi standard crt:\
Then type:
# vi /usr/lib/terminfo/terminfo.src
Find the line that reads:
ansi|ansic|ansi-437|ansi80x25|Ansi standard console,
and change it to read:
ansi|ansic|ansi-437|ansi80x25|scoansi|Ansi standard console,
After adding the scoansi definition to the terminfo.src file, run
the terminal info compiler:
# tic (no relink/reboot necessary)
SCO has added "scoansi" as an official terminal type (or, more precisely, a
preinstalled synonym for what it now calls "ansi") in OpenServer 5.0.5 and
Unixware 7.0.
Also see Bug 417.
It has been noted that the \$(xxx) construction for accessing environment
variables is not case sensitive, as it should be. This is a bug or a feature
of Windows 95 inherited from DOS (Kermit gives the environment variable name
literally to Windows, but Windows ignores the case of the letters).
When running K95 host mode in Windows 95 (but not NT or OS/2), callers could
experience fractured screens, lost or duplicated data, etc. The problem was
traced to problems in Windows 95 overlapped serial i/o, so the HOSTMDM.KSC
script was modified to not use it in K95 1.1.12. Similarly, keystrokes sent
to the hostmode screen could be lost if they were too close together. This
was caused by a bug in HOST.KSC, also fixed in 1.1.12.
If you specify a password on the login page of a Dialer entry and find that
the connection is not made at all, try removing the password (and specifying
NONE for the prompt). If that fixes the problem, but you still want to
specify a password (even though it is a security risk to store passwords on
your computer), then try changing the password on the host computer (e.g. by
appending an X to it) and entering the new password into the Dialer entry.
As of 1.1.14 you can also ask that the Dialer prompt you for the password
each time you want to connect to the host.
The problem is fixed in version 1.1.16.
(Fixed.)
(Fixed.)
There is no way specify a scrollback buffer size of 0, or for that matter,
less than 512 lines. Thus, SET TERM SCROLLBACK 0 can't be used to disable
scrollback. Instead, assign \Kignore (or any other non-scrollback related
verb, etc) to the scrollback keys; for example:
set key \4385 \Kignore ; Ignore PageUp
set key \5409 \Kignore ; Ignore Ctrl-PageUp
set key \4388 \Kignore ; Ignore the Home key
and so on.
(Fixed.)
In VT220 and VT320 emulation, the PC's numeric keypad is mapped by default
to the VT terminal's numeric keypad. Prior to version 1.1.10, however, this
was not true for VT100 and VT102. But the numeric keypad functions are just
as important in VT100/102 emulation as they are in VT220/320, and a lot of
users of VT100/102 emulation complained that the numeric keypad didn't work
right. So in version 1.1.10, the VT100/102 numeric keypad was mapped the
same as the VT220/320 numeric keypad.
However, this change (like all changes) surprised other people who had become
accustomed to the previous VT100/102 mapping, which was the same as the
mapping for the editing and arrow keys with the same labels.
If you want to restore the previous mapping, add the following to your
K95CUSTOM.INI file:
SET TERMINAL KEY VT100 \290 \Kdnscn
SET TERMINAL KEY VT100 \291 \Kendscn
SET TERMINAL KEY VT100 \292 \Khomscn
SET TERMINAL KEY VT100 \293 \Klfarr
SET TERMINAL KEY VT100 \294 \Kuparr
SET TERMINAL KEY VT100 \295 \Krtarr
SET TERMINAL KEY VT100 \296 \Kdnarr
SET TERMINAL KEY VT100 \302 \{127}
SET TERMINAL KEY VT100 \400 \Kignore
SET TERMINAL KEY VT100 \2411 {-}
SET TERMINAL KEY VT100 \4143 /
SET TERMINAL KEY VT100 \4365 \{13}
SET TERMINAL KEY VT100 \4463 /
SET TERMINAL KEY VT100 \4496 \Kignore
SET TERMINAL KEY VT100 \4543 /
Replace VT100 with VT102 or other terminal type that you want to have these
definitions on the numeric keypad. (A copy of this file is included with
Kermit 95 1.1.11 and later as KEYMAPS\KEYPAD.INI.)
(Fixed.)
(Fixed.)
The ZMODEM ZINIT packet, by which Kermit recognizes the beginning of a ZMODEM
download, is indistinguishable from a Wyse "Enter Controller Print" sequence.
The terminal emulator takes precedence. Similar types of confusion occur
with Volker Craig and Data General terminal types. This is a conflict, pure
and simple. Kermit 95 1.1.11 adds some new SET TERM AUTODOWNLOAD options to
address such conflicts; type HELP SET TERMINAL at the K95 prompt
to see them.
Don't expect default Kermit 95 mouse button assignments to work as described
on pages 55-58 of the Kermit 95 manual if you have installed a nonstandard
mouse driver and/or remapped your mouse buttons with a product such as the
Clicker program that comes with the Logitech Trackman. For example, if you
have used Clicker to translate Mouse Button 2 into Double-Left-Click, then
don't expect double-clicking button 2 to paste from the Clipboard to the host.
Work around such situations by using the SET MOUSE command to remap mouse
events as desired, e.g.
SET MOUSE BUTTON 3 NONE DOUBLE-CLICK \Kpaste
(Fixed.)
(Fixed.) But we don't need the
WP51.INI file any more since K95 now has its own built-in Word
Perfect Keyboard Mode.
(Fixed.)
"I'm running K95 1.1.10 under Windows 95. If I invoke the DOS EDIT program
from the Kermit command screen either with !edit or by pushing to DOS and
invoking edit from DOS, when I return to K95 anything which is supposed to
display with a yellow background displays instead as blinking on a brown
background. Normally, my 'select' colors are black on yellow. If I am in the
command screen and type 'show terminal', the colors display properly. If I
then type '!edit' and exit edit and return to the K95 command screen, the
previously displayed black on yellow is now blinking black on brown. Any
colors which I change to yellow background become blinking on brown instead.
This is a limitation of Microsoft Windows 95 and NT 16-bit DOS support.
16-bit DOS programs, such as EDIT.COM, are allowed to toggle the
meaning of the VGA background-intensity bit. Instead of using it to allow
bright colors as the background, EDIT.COM wants to use it to blink
the text. The use of this bit to represent blinking is not supported in
window sessions but is supported in full-screen sessions. There is no
mechanism by which a 32-bit Windows application such as Kermit-95 can change
the meaning of this bit.
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
OS/2 C-Kermit 5A(191) did not allow all key combinations to be mapped. Even
worse was the fact that several of the combinations would generate the same
keycode. When Kermit-95 was being planned it was decided that it would allow
all keyboard combinations to be mapped to unique values. Meeting this goal
required discarding the entire OS/2 C-Kermit keymap table and starting from
scratch.
Because of Kermit/2's more intensive probing of the keyboard than prior
OS/2 Kermit versions, it no longer works with SWAPDCP (a public domain,
use-at-your-own-risk method of swapping Ctrl and Caps Lock in OS/2). See
the OS/2 Notes for details.
In May 1997 hackers discovered that any Windows 95 or NT system could be
crashed by sending TCP Out Of Band (OOB) data to it on the NETBEUI port.
Microsoft's reaction was to publish a "hotfix" for Windows NT that disabled
processing of OOB data by Winsock; refer to Microsoft Knowledge Base Document
Q143478 [winnt], "Stop 0A in TCPIP.SYS When Receiving Out Of Band (OOB) Data".
But RLOGIN protocol works by using OOB data, and therefore if you have
installed these fixes, you will not be able to use RLOGIN. This situation was
addressed in a second "hotfix" issued May 24. Each fix is simply a
replacement of the TCP/IP stack (TCPIP.SYS). You can (and should) install the
latest one whether or not you installed the first one. As of this writing
there is no fix from Microsoft for Windows 95.
NOTE: If Windows NT 4.0 Service Pack 3 has been installed, the Post-SP3 OOB
fix must be installed, since SP3 contains the first version of the OOB fix.
(Fixed.)
(Oops, this is a duplicate of Bug #224.)
(Fixed.)
In Windows 95/98 and NT it is possible to install more than one
keyboard layout in the Keyboard Control Panel on the Input Locales
page. When more than one Input Locale/Layout combination is defined,
you are required to choose one of them as the default, and to choose a
key combination (such as Left Alt + Shift) to switch among them. This
is called the Switch-Locale key sequence. The default Input-Locale is
the Input-Locale that each Window uses when it is opened.
For instance, it is possible to install:
Input Locale Layout
------------------------------------------ ---------------
English (United States) US
Russian Russian
with "English (United States)" as the default input locale and a
Switch-Locale key sequence of "Left_Alt+Shift". Since the keyboard
only has one alphabet on it, this "alphabet shift" key is needed to
switch between the two defined alphabets. In Russian Windows, for
example, it is necessary to type Roman letters when you type Windows
file names or Kermit 95 commands and Cyrillic letters when writing in
the Russian language.
Although the behavior is supposed to be the same on Windows 95/98 and
Windows NT, it turns out that it is not.
In K95 on NT you can give the following commands:
SET TERMINAL CODE-PAGE 1251
SET TERMINAL LOCAL-CHARACTER-SET CP1251
SET TERMINAL REMOTE-CHARACTER-SET CP866
When the current locale is "English (United States)" K95 receives
Roman characters from the keyboard. When the current locale is "Russian"
K95 receives Cyrillic characters from the keyboard. To toggle between
the two press the Switch-Locale key sequence.
When issuing commands to K95 switch to the "English" locale and when
sending Cyrillic characters to the host switch to the "Russian"
locale. There is no limitation to mixing Roman and Cyrillic
characters via this method at the K95 command prompt or in terminal
mode.
Note that only characters that are in the active code page may be delivered
to K95 from the keyboard. In the above example code page 1251 is
used because it is available on all versions of NT and it contains all
of the Roman characters needed for U.S. English and the Cyrillic
characters used by CP866. (CP866 is not available on the U.S. version
of NT but may be available on some international versions.)
However, when using Windows 95/98 locale switching does not work
properly in 32-bit console applications. (The following was
tested on Windows 95 OSR2 - Russian release.) The Russian Win95 uses
CP866 as its default code page for the DOS and Win32 Console
environments. Therefore, both Roman and Cyrillic characters are
available for display.
For DOS applications the current locale is ignored. All keystrokes
generate Roman letters regardless of what locales are installed.
For 32-bit Console applications when both the "English (United
States)" and "Russian" input locales are defined and a Switch-Locale
key sequence specified it is not possible to switch the input locale
from the Default (as selected in the Keyboard Control Panel.) The
result is that only the default language may be used. If the default
locale is "Russian" then only Cyrillic characters may be generated
which makes it impossible to type the English commands necessary to
control Kermit 95. If the default locale is "English" then it is not
possible to generate Cyrillic characters using the Windows Keyboard
layouts.
In this case it is possible to use K95's built-in English-to-Russian
keyboard mode with the following limitations:
- Your default Input Locale must be set to "English".
- CapsLock cannot affect the punctuation marks that are used to
generate some of the Cyrillic letters.
- it can not be used at the K95 command prompt. Therefore, file names
containing Cyrillic characters cannot be specified for file transfer
by typing. A work around is to list the files with the DIR command
and then copy/paste the file names with the mouse.
NOTE: As of 1.1.17, K95 implements its only Keyboard Layout management
routines (on Windows 95) such that when K95 starts it will use the
Default Layout. Each press of the LeftAlt-Shift or Ctrl-Shift
combinations will toggle the Keyboard Layout.
(Fixed.)
(Fixed.)
The Kermit 95 version of TELNET.EXE (which simply starts Kermit 95)
assumes that K95.EXE is in the PATH. If it is not, and it is also
not in your current directory, then TELNET.EXE will fail to start
K95.EXE. This happens most frequently when Kermit 95's
TELNET.EXE is specified as the Telnet application for Netscape or
other Web browser. Solution: Add the Kermit 95 directory to your PATH, or
else try specifying "C:\K95\K95.EXE -J" as your Telnet
application, replacing "C:\K95" by the actual Kermit 95 disk and directory.
Also see Bug 111. Fixed in 1.1.16 as long as
TELNET.EXE is in the same directory as K95.EXE. Ditto for
RLOGIN.EXE.
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
You don't, at least not for now. SSH uses
encryption algorithms that fall under one or more of the following: are
prorietary, patented, require licensing, and/or are regulated for export by
the US Government. They cannot be implemented in Kermit products without
either significant increases in the price of the software, maintenance of
separate versions for US and non-US markets, and halting distribution of
source code (for C-Kermit), binaries, and patches via the Web.
Kermit 95 1.1.16-1.1.19 have added support for Kerberos, SSL/TLS, and other
forms of security. SSH might come later, after its patents expire.
In Windows 95, the available font sizes are listed in the font-size dropdown
list in the console window toolbar. The repertoire of the font is the current
console code page, such as CP437, CP850, CP862, etc. See Item 5 for further
information.
In Windows NT, you have a wider selection of fonts, including Unicode fonts
(which are not limited to the repertoire of the current code page), but you
still can't use just any font in the system; only the ones listed in the Fonts
section of console window Properties dialog. NT includes and excludes fonts
from this list based on spacing, proportionality, height/width ratio, encoding,
and other factors. The GUI version of K95 will not suffer the same
constraints.
(Fixed.)
Suppose you use a special prefix to dial toll calls (long-distance calls),
as described on pages 112-113 of "Using C-Kermit", 2nd ed., for example:
set dial ld-prefix 18005551234
If you do this, make sure you also have specified the toll-free area codes of
your local country, e.g. 800 and 888 (and soon, also 877) in the North
American Numbering Plan (NANP). Otherwise your toll-free call will look like
a toll call to Kermit, and the long-distance prefix will be applied, which in
some cases results in failure to complete the call (e.g. you get a recorded
message like "Please don't use an access code to dial a toll-free call").
Also, make sure your toll-free dialing prefix (at least in the NANP) is the
normal long-distance dialing prefix ("1" in the NANP), not the access code for
your long-distance service provider.
Prior to version 1.1.14, Kermit had no special provision for "blind dialing",
i.e. forcing the modem to dial even when it does not recognize a dial tone, as
is often necessary with PBXs, in hotels, when travelling, etc, so to handle
this you had to change your modem's init string (this section explained how).
Blind dialing is now handled by the new SET DIAL IGNORE-DIALTONE command.
See Section 2.1.4 of the C-Kermit 7.0
update notes.
You can't open the same log file in two (or more) different K95 sessions
and have them both write to the same file. So if you want to create a
session log, or transaction log, etc, from multiple K95 sessions at the same
time, give each one a different name. Otherwise, the LOG TRANSACTIONS
command that you give to the second session will destroy the log created by
the first session (and so on).
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
The RUN command does not parse filenames, because the command to be run might
not be a file -- it might be built in to your shell. So in the RUN command
(and its synonyms, "!", "SPAWN", etc), normal Kermit parsing rules regarding
backslash apply. So to include any directory separators in a RUN or similar
command, you must double them: "RUN \\FOO\\BAR\\BAZ.BAT". Note that in this
case you can't use forward slashes as directory separators because the normal
Windows command parser does not allow them, even though the operating system
itself does (which is why you can tell K95 to "SEND /FOO/BAR/BAZ.BAT", etc).
Furthermore, in some cases you must also use double backslashes as directory
separators in any filenames that are used as operands to the command. (But
most programs other than the standard Windows command shell recognize forward
slashes...) Well, this is pretty complicated, so here are some examples:
run c:\windows\notepad.exe d:\k95\tmp\oofa.txt - doesn't work
run c:/windows/notepad.exe d:/k95/tmp/oofa.txt - doesn't work
run c:\\windows\\notepad.exe d:\\k95\\tmp\\oofa.txt - works
run c:\\windows\\notepad.exe d:/k95/tmp/oofa.txt - works
In the last example, we use double backslashes in the command name, which are
converted by Kermit to single backslashes and then passed the Windows command
shell, so the command shell sees "c:\windows\notepad.exe
d:/k95/tmp/oofa.txt".
So it starts Notepad and passes "d:/k95/tmp/oofa.txt" to it as an
argument. Notepad uses the regular file-opening APIs, and so it has no
trouble with forward slash as a directory separator.
Of course you can always take the easy way out if the RUN command contains no
Kermit variables or backslash escapes, as follows:
set command quoting off
run c:\windows\notepad.exe d:\k95\tmp\oofa.txt - works
run c:\\windows\\notepad.exe d:\\k95\\tmp\\oofa.txt - doesn't work
set command quoting on
SET COMMAND QUOTING OFF disables special handling of backslash by Kermit's
command parser.
Here's a case study, illustrating the effect of multiple levels of backslash
quoting... Suppose you want to:
- Transfer a ZIP file from a UNIX workstation to Kermit 95
- Unzip the file with the unzipped files being placed on a floppy disk
- Change the attributes of the unzipped files on the floppy disk
- Remove the transferred ZIP file from the PC
All under the control of the remote C-Kermit. This can be done with an APC
command; you have to "set terminal apc unchecked" in K95 for this to work.
Please read about this in the "Using C-Kermit" manual (2nd Ed). The following
shell script does the trick:
#!/bin/sh
if [ $# = 0 ]; then
echo "usage: $0 filename.zip"
exit 1
fi
kermit -YQ -C \
"set macro error on, binary, move $1,\
apc {run unzip $1 -d A:/,del $1,run attrib +R A:\\\\\\\\*.*},exit"
$1 is the shell command argument, like \%1 in Kermit.
We're unzipping the ZIP file into the top-level directory of the A: drive.
UNZIP understands forward slash as a directory separator, so "A:/" is
OK as an operand to UNZIP.
But in DOS commands like ATTRIB, we have to use "\". But this is also (1) the
shell's escape character, (2) C-Kermit's escape character, and (3) Kermit 95's
escape character. So when telling the shell to tell C-Kermit to tell K95 to
run a command that refers to a directory, we have to include EIGHT (= 2 to the
3rd power) backslashes for each ONE that we actually want passed to the DOS
command.
If you include a /K switch in a RUN command, this is a switch to the Windows
command processor that means "Don't return" (/K = "keep"). So it doesn't
return. There is nothing K95 can do about this. This can happen
inadvertently if you were trying to use forward slashes as directory
separators: "run /k95/k95cinit". Again: you must use backslashes as
directory separators in program names given to the RUN command: "run
\\k95\\k95cinit" (or "set command quoting off" and single backslashes).
(Fixed.)
(Fixed.)
CTL3D32.DLL is the Microsoft library that contains the 3D controls used by
the Dialer. The Dialer can't run without it. This DLL is a critical
component of Windows 95 that Microsoft forgot to include in Windows 95, and
so it has to be packaged with *each* GUI application, and then either
installed or not installed in this or that place depending on all sorts of
things -- and to make matters worse, there are multiple incompatible
versions. The same is true on NT, apparently. Microsoft itself is silent on
the subject, but there are entire web sites devoted to just this topic (try
an Alta Vista search on "ctl3d32.dll"). For 1.1.12, we have made copies of
this DLL for Windows 95 and NT (separate copies, of course) available on our
website. In 1.1.13 and later, it is included in the release and patches. A
new program, CTL3DINS.EXE, distributed with 1.1.13 and later, uses the
Microsoft VerInstallFile() service to install the appropriate DLL (Window 95
versus NT; Intel vs Alpha, etc) if necessary (e.g. if it is missing on your
PC or your version is older). The behavior of VerInstallFile() is exactly
what Microsoft provides, no more and no less (as to messages, actions, lack
of choices, failure to back up any old copies).
The command line for CTL3DINS.EXE accepts and optional /F parameter
which means "Forcefully install CTL3D32.DLL". This might be needed
if a previous application installed the NT version on Win95 or the Win95
version on NT, or if the previous version is write-protected, or at any other
time the CTL3DINS program states that -F might be used to override an error
condition.
If you choose a TAPI device as your default communications device in SETUP, or
if you have "SET LINE TAPI" or "SET TAPI LINE" command in your K95CUSTOM.INI
file, then when you start K95, it opens the TAPI line; this can take a long
time (it's TAPI, not Kermit, that is slow). If you try to make a network
connection or to exit from K95 after the TAPI device is open, then K95 has to
ask TAPI to release the line; this takes a long time too. So if you will not
be using a TAPI device as your normal communications device -- for example, if
you use K95 more frequently for network connections than for dialing out --
remove the aforementioned commands from your K95CUSTOM.INI file.
Then when actually do want to dial out from K95.EXE, you'll need to
"set tapi line" first (but you won't need to do anything special if you make
connections from the Dialer).
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
When using the Paste button on the Windows toolbar (which works only if you
follow the hints in item 61 in this file), and using VT100 or higher
emulation, K95 sends Esc O P before and after the pasted text. This is
because Windows stuffs a Num Lock key-press code into the keyboard buffer
before and after the paste. There is no way for K95 to tell whether these Num
Lock keycodes came from the user pressing the key, or Windows stuffing the
keypress into the buffer. So K95 sends whatever is associated with Num Lock,
which, in the case of VT emulation, is normally the PF1 or "Gold" key code,
Esc O P, and perhaps other sequences in other emulations. The only way to
defeat this behavior is to tell K95 to ignore the Num Lock key:
set key \400 \Kignore ; Windows 95
This problem does not occur in OS/2.
The reverse writing direction feature of the VT320 (DECRLM, CSI ? 34 h / l)
is not supported. Thus, Hebrew terminal emulation works only when the host
application positions the cursor in the right place to write each character.
This is the case (for example) with the ALEPH bibliographic system. Reverse
screen writing will be added in a future release.
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
In Windows 95 and NT, URL hot spots (and the BROWSE command) invoke your
browser via ShellExecute() if you SET your BROWSER to nothing.
But this is not the default browser
setting since it tends not to work. One reason for this has been identified:
When the browser is installed, it is supposed to (a) install URL.DLL, and (b)
add URL.DLL to the Registry key:
HKEY_LOCAL_MACHINE\Software\Microsoft\Windows\CurrentVersion\Explorer\
ShellExecuteHooks
Some browsers, however, register URL.DLL in HKEY_CURRENT_USER and/or
HKEY_USERS rather than HKEY_LOCAL_MACHINE. In that case, ShellExecute() can
not find URL.DLL, and therefore can not handle URLs. This has been observed
(for example) with various versions of Netscape, such as 2.0. If you install
Microsoft Internet Explorer, ShellExecute() will work, even if you do not make
Explorer your default browser. Now we know why: because it registers URL.DLL
under the same key used by ShellExecute() to locate it.
(Fixed.)
(Fixed.)
When you activate a Window, if the current state of the NumLock, ScrollLock,
CapsLock, ... flags for that window do not match those of the other sessions,
it places a NumLock, ScrollLock, CapsLock, ... key event into the message
queue for that window.
If NumLock is On, and you use the Toolbar Paste Button to paste text into the
Window, Windows 95 will toggle NumLock off and then on by surrounding the text
with NumLock key events. See Bug Number 356.
If this is a problem for you, the only thing that you can do is not assign
Kverbs to those keys. This means \Kholdscrn and \Kpf1 cannot be assigned to
the ScrollLock and Numlock keys. Instead you should assign \Kignore.
This is a limitation of the Windows 95 Console environment. It does not
affect GUI applications (as far as we can tell) nor does it occur on Windows
NT.
(Not a bug) In Windows NT, it is not possible to "set port lpt1" -- i.e. to
use a parallel port as a communications device, because bidirectional parallel
ports are not supported by the NT drivers. But they are in Windows 95, which
includes a COM-to-LPT interface. So in Windows 95, you can:
SET PORT LPT1
SET PRINTER /DOS-DEVICE: LPT1
(or other parallel-port device). Only DOS names are allowed. Bidirectional
parallel port support is under development by Parallel Technologies
(http://www.lpt.com) for inclusion in Windows NT 5.0 and possibly for
retroactive distribution.
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
SCO UNIX, OpenServer, etc, and IBM AIX allow their console and/or terminal
drivers to be configured to translate character sets for you. DON'T DO THIS
WHEN USING KERMIT! First of all, you don't need it -- Kermit itself already
does this for you. And second, it will (a) probably ruin the formatting of
your screens (depending on which emulation you are using); and (b) interfere
with all sorts of other things -- legibility of non-ASCII text on the terminal
screen, file transfer, etc. On SCO systems, use:
mapchan -n
to turn off this feature. On AIX use:
setmaps -t NOMAP
or:
chdev -l tty??? -a imap='none' -a omap-'none'
(replacing ??? by your tty number)
WinPrinters (printers that only work with Windows 95) cannot print plain text.
But Kermit 95 prints using "raw mode" because many host applications send
printer-specific files such as Postscript and PCL. Meanwhile, the Windows
32-bit printer drivers do not allow plain text to be sent to this type of
printer, but rather expect the application to convert the plain text to dots.
Workaround: SET PRINTER to a file, and then use the DOS "PRINT" command (or
"COPY /B filename PRN" or even "notepad /p filename") -- oddly enough, these
printers do come with 16-bit drivers that do the needed conversions;
unfortunately, appropriate conversions are not provided automatically for
32-bit applications.
Clients that don't support the minimal ANSI set of screen clearing,
line-clearing, and cursor positioning sequences will not be able to view
the K95 host menu screens correctly. To be addressed in a future release
of K95 hostmode.
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
Sometimes K95 crashes with a message something like "Winoldap: This program
has performed an illegal operation and will be shut down. If the problem
persists, contact the program vendor." If you look at the message in more
detail, you will usually see that the crash occurred not in K95 at all, but in
the Windows kernel. Furthermore, please be aware that "Winoldap" does not
refer to K95 -- it is not an "old" application, a DOS application, a 16-bit
application, nor is it "old" in any other way. In fact, this term refers to
the Microsoft Console window in which Windows 95 forces K95.EXE to
run.
(Fixed.)
(Fixed.)
(Fixed.)
If you have a 7-bit National Replacement Character Set (such as Swedish,
German, etc -- the ones with "national" names) as GL and an 8-bit display-only
character set, such as DEC Special Graphics or DEC Technical, as GR (as shown
by "show character-sets"), you won't be able to type your language-specific
characters as expected. Workaround: See item 477.
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
(Fixed.)
At least the following problems showed up in 1.1.14 and 1.1.15:
- Failure to update sections of the screen during terminal emulation;
- Coloration mixups;
- In Wyse emulation, entry into protected mode caused an infinite loop;
- VT scrolling region problems;
- GET would fail after an XMODEM SEND.
These were traced to bugs in the optimizer of the new release of the Microsoft
Visual C/C++ compiler SP2 that was used to build Kermit 95 1.1.15 for Windows.
Rebuilding the program without optimization makes these problems disappear.
K95 1.1.16 was built with SP3, with optimization turned off. Thus the
K95.EXE file is considerably bigger, but the risks are lower (the
optimization bugs in 1.1.15 were quite subtle and took a good while to show up
and to be properly diagnosed).
The select() call problems appear to have been fixed in build
4.0.1656 of windows\system\wsock32.dll. This release is currently available
for download at the web site (the next two lines should be concatenated):
http://www.microsoft.com/windows/downloads/contents/updates/w95sockets2/default.asp
With the select() bug fixed performance is only slightly worse with
Winsock 2.2 as it is with 1.1. However, it should be noted that all
development on 1.1 has now officially been stopped by Microsoft and
all future fixes will be delivered in new builds of 2.2. This will
include an upcoming fix to the Domain Name resolution bugs in the
current 1.1 and 2.2 code.
The Dialup Networking 1.2 and DCOM'95 updates must be installed prior to
Winsock 2 (if they are to be installed at all):
http://www.microsoft.com/windows/downloads/contents/updates/W95DialUpNetw/default.asp
http://www.microsoft.com/com/dcom95/download-f.htm
To check the version number of a DLL, open the \WINDOWS\SYSTEM\ folder
and then right mouse click the desired DLL and choose Properties from
the Menu. The Version information is available on the "Version" page.
Why can't I redirect K95's standard output? Because K95 does not write to
standard output, it writes directly to the screen device, which can not be
redirected. Why does it do that? See items 46 and 195. Fixed in 1.1.16.
"This program has performed an illegal operation and will be shut down" can
happen when Windows 95 runs out of swap space. Windows conveniently blames
the problem on the application. You can reconfigure your swapfile, e.g. by
moving it to a disk with more space, in the Control Panel (search for "Swap"
in Help).
One user reported the following: While running K95, every so often the disk
goes crazy, the PC slows down, and the fan starts roaring. Diagnosis:
coincidence, nothing to do with Kermit; it was Microsoft Office 97 Fast Find,
which periodically wakes up and searches the disk(s) to build indexes.
We've heard of WinModems, WinPrinters, and even WinCPUs, but a WinFan???
Note: the same effect can result from any number of other helpful utilities,
such as disk defragmenters, etc, and if your PC is connected to the Internet,
also from people trying to break (into) it.
In Windows NT (but not 95) SET PRINTER /WINDOWS-QUEUE did not select the
default printer if it was a network printer; SET PRINTER /WINDOWS-QUEUE:?
showed only locally connected printers. Fixed in 1.1.16.
In Windows 95, K95 was inadvertantly changing the Console from text to binary
mode. No problem under most circumstances unless the program that is starting
K95 under Windows 95 relies on text-mode for gets() to succeed. Fixed in
1.1.16.
Under certain circumstances, K95 or Windows 95 itself can appear to freeze
when you try to exit from K95. This has been observed when using Option 11 to
exit from the HOSTMODE script in Windows 95. This problem is fixed in 1.1.16.
HINT: Remember not to try to exit from K95.EXE by clicking the Close
[x] box on the menu bar except when K95 will not exit normally. When you
click the Close box, Windows closes Kermit 95 without releasing its resources,
which can lead to all sorts of problems until the next time you reboot
Windows. This is a Microsoft feature over which Kermit has no control
whatever, since Windows does not send it a Close message when you click this
box, nor does it give K95 any way to disable this box.
UPDATE: In version 1.1.19, K95 disables the use of [x] box. If you need to
shut down Kermit 95 abnormally, select the K95 window and then use
Ctrl-Alt-Del.
An Exit Warning was generated if a TAPI device was open even if a call was not
in progress. Fixed in 1.1.16.
The RUN (!) command did not properly return the exit status of the command
or program that it ran, so it could be tested for success or failure, and
did not set the \v(pexitstat) variable to the program's or command's return
code. Fixed in 1.1.16. However, note that this feature works only if the
underlying shell cooperates. The standard Windows 95 shell, COMMAND.COM,
does not. Thus in Windows 95, K95's RUN command will always succeed when
COMMAND.COM is the shell. Proper status is returned by the Windows NT version
of COMMAND.COM, as well as by most alternative shells for Windows 95.
Fixed in 1.1.16, except there still can be problems in this and other
commands completing filenames that contain spaces.
Ditto for the \ffiles() function. Incidentally, the DIRECTORY
command is implemented internally by calling the \ffiles() function,
and so this would destroy the user's \fnextfile() list if one was
active. All these problems are fixed in 1.1.16.
Bug, fixed in 1.1.16.
The SET TITLE command did not take effect until the next CONNECT command.
In 1.1.16, it takes effect immediately.
A BEEP command might take a long time to work if you have not used any
sounds recently; sound driver code might be swapped out or not yet loaded,
in which case it must be read in, and then the WAV file for the desired
sound must be read, and then the sound must be generated, which might take
some time.
The CONNECT command should refuse to work if a TAPI device is open but no
call is in progress. Fixed in 1.1.16.
A TAPI LINE device would not be closed properly when released (e.g. when K95
was told to hang it up and switch to another device), and this would prevent
other applications from accessing the device until K95 terminated. Fixed in
1.1.16.
The SET TCP NODELAY command was controlling the wrong socket parameter,
the DEBUG capability instead of the Nagle algorithm. SET TCP commands were
not gracefully recovering from setsockopt() call failures.
Fixed in 1.1.16.
The \v(ipaddr) variable did not return the IP address of your PC
until after you had made a TCP/IP connection. Fixed in 1.1.16. But note that
this variable still might not return the IP Address that you expect because of
virtual adapters, etc; \v(ipaddr) is set to the IP address of the
current connection each time a new connection is made.
Echoing, binary modes, NAWS, and most other Telnet options would not be reset
properly when connecting to ports other than 23 (the Telnet port). This could
result in Telnet negotiations failing to complete because Kermit thought the
negotiation had already been performed. Fixed in 1.1.16.
The TELNET Network Virtual Terminal specification requires that a Carriage
Return (CR) that is not followed by a linefeed (LF) must be followed by a
NUL (ASCII 0). These NULs should be stripped away before processing incoming
data; in most cases they were. Exceptions: (1) they were recorded in the
session log, and (2) they were not ignored by the INPUT command. Both
exceptions fixed in 1.1.16
"rlogin://host" would not be recognized as a valid URL, since "rlogin" is not
a valid TCP service name (the proper name is "login"). Not exactly a bug,
but "rlogin" is recognized as a synonym for "login" in 1.1.16 and later.
Wyse graphic characters were erroneously converted to the local Code Page
instead of to Unicode in Windows NT. Fixed in 1.1.16.
Bug, fixed in 1.1.16.
DG transparent printing would fail to send the escape character to the printer
if the escape sequence was not the print terminator. Fixed in 1.1.16.
NEL (= Esc E) did not scroll on bottom line of screen or scrolling region
in terminal types that followed ANSI X3.64 (including VT220 and 320, Wyse
370, ...) Fixed in 1.1.16.
<CSI> Pn X was not erasing characters properly; it was writing blanks with the
current attributes, rather than with no attributes. Fixed in 1.1.16.
In SCOANSI and AT386 emulation, attempts to program function keys were being
misinterpreted as Privacy Messages (PM). Fixed in 1.1.16.
Status Line Messages were not being parsed; unrecognized escape sequences were
being displayed on screen. The entire emulation was rewritten to parse HP
sequences in a general way, avoiding special cases.
BA80 emulation did not handle Function Key Labels or Status Line commands.
Fixed in 1.1.16.
When Terminal screen is in debug mode and a CR and/or LF arrives when the
cursor is on the right margin, an extraneous blank line in debug color is
displayed. Fixed in 1.1.16.
They don't, because AIXTERM remaps these as extended function keys; for
example, Alt-x sends <CSI>102q (see DEFAULT.INI). K95 reassigns its own
functions to the uppercase Alt keys, Alt-X, Alt-H, etc.
Mouse-pasting using the Kermit method (double-click button 2) tends not to
work because any miniscule motion between the two mouse clicks raises a drag
event, thus putting Kermit into Mark Mode (which can be canceled with a
terminal reset (Alt-R) or with Ctrl-F2 (Cancel Mark Mode). Fixed in 1.1.16.
Workaround - Add the following commands to your K95CUSTOM.INI file:
SET MOUSE BUTTON 2 NONE DRAG \Kmousecopyclip
SET MOUSE BUTTON 3 NONE DRAG \Kmousecopyclip
On certain PCs, under certain circumstances, there might be a slight delay in
echoing of the characters you type. A large part of the delay is in the
connection itself, but other components of it are attributable to Windows
scheduling of K95's several threads and its semaphore management, especially
under heavy system load, and to overhead from CONAGENT.EXE, which
controls the window that K95 runs in. The situation was improved dramatically
in K95 1.1.16.
Note that you can use SET TERMINAL SCREEN-UPDATE { FAST, SMOOTH } to control
the tradeoff between high throughput (FAST) and snappy echoing (SMOOTH); see
the SET TERMINAL Command List
for details. Also see item 27 in this file.
Bug, fixed in 1.1.16.
A SET TERMINAL TYPE command in an APC command would reset the terminal
emulator, thus clearing the APC status even though an APC was still in
progress. Fixed in 1.1.16.
On Windows NT or OS/2, if K95 was set to FILE NAMES LITERAL and a file
arrived to be stored on a FAT partition restricted to 8.3 filenames,
and the file had a long name or a name otherwise illegal for FAT, the file
transfer would fail unnecessarily. In 1.1.16, the name is transformed just
enough to make it legal for FAT.
Kermit or Zmodem autodownloads would fail to work if K95's parity was set
differently to what the host was sending. Fixed in 1.1.16.
Due to some reorganization of K95's threads for better error and interrupt
trapping, K95's file-transfer display thermometer might not go all the way to
the end on the last or only file, especially if it is a short file. However,
the Percent Done field still registers 100%, and the Last Message field still
shows file and byte count and characters per second, and indicates a
successful transfer (if the transfer was indeed successful). SHOW STATUS will
also reveal the success or failure of the transfer. Fixed in 1.1.16.
After downloading when XMODEM, YMODEM, or ZMODEM, K95 would report twice the
number of files that were actually downloaded. Fixed in 1.1.16.
An internal problem that occurred only when the terminal type was set to
ANSI-BBS, SCOANSI, and AT386 emulations, which don't use 7-bit National
Replacement Character-sets (NRCs) anyway. Fixed in 1.1.16.
Refer to entry 430. Fixed in 1.1.16. Characters can now be assigned to keys
without translation with the new SET TERM KEY /LITERAL switch.
When you choose HP-Roman8 as the remote character set in the Dialer, it writes
"set term remote-char hproman8", but it should have included a dash
("hp-roman8"). Fixed in 1.1.16.
Bug, fixed in 1.1.16.
The number of files reported by \ffiles() was always at least two
greater than it should have been, since it was counting "." and
".." (and any other directory) as files. Fixed in 1.1.16.
REMOTE QUERY KERMIT function(\%a) did not evaluate its argument before sending
the query to the server; thus it was not possible to have the server run the
function with client-side variables as arguments. Fixed in 1.1.16. Details
in Section 5.2 of the C-Kermit 7.0
Update Notes.
When Kermit 95 receives an incoming TCP connection, it looks up the incoming
IP address to get the hostname, primarily so that your hostmode logs and
displays will contain accurate information and not spoofed or forged names;
this is standard practice on the Internet these days. This query, however,
could take a long time -- if your local Domain Name Server can not satisfy it,
it goes out to the Internet, and we all know how snappy the Internet at
times... In 1.1.16, you can defeat the pause with the new SET TCP
REVERSE-DNS-LOOKUP OFF command.
(Duplicate of 419; see Bug Number 419.)
Typographical error in hostmode.ksc, fixed in 1.1.16.
K2DIAL.EXE would lose data in the Settings Notebook Login Script Text
box if the Printer entry field is edited AND the LOGIN page is not displayed.
The bug appears in both Warp 3.0 and 4.0. Diagnosis: Compiler optimization
bug in IBM VAC++. Fixed in 1.1.16.
The Dialer Printer Type Combo box could not be set to File, Pipe, or None
types from saved values. Fixed in 1.1.16.
The Dialer entry Communications page would not retain the override speed
unless flow control was also overridden. Fixed in 1.1.16.
(Not a bug.) If you can't get the Alpha paging script to work:
- Check with your paging service to make sure you have the right
settings (phone number, speed, parity, flow control), and modify the
APAGE macro in APAGE.KSC accordingly.
- If, when dialing, you still get only garbage and/or immediate hangups,
you'll need to identify the settings to *force* your modem to dial the
paging service in such a way that the two modems can communicate. Look
in your modem manual for commands related to modulation, fallback, and
speed buffering. You want the modem to dial at 1200 (or whatever the
appropriate speed is), using the appropriate modulation, and you want
it to stay at that speed after the connection is made, and you want the
interface to stay at that speed too. It often turns out that old
modems like the Hayes 1200 work just fine with paging services, where
more modern models tend to fail because of complications during
the modulation and protocol negotiation phase -- so do whatever you can
to turn your fancy new modem into a dumb old one. Hint: use the new
SET MODEM PREDIAL-INIT command to specify any commands you want sent to
the modem just prior to dialing.
- In the APAGE macro, try changing the flow control and/or parity to
NONE, even if that is contrary to what your paging service told you.
- Try adding "set output pacing 200" (or other number) to the APAGE
macro -- this makes the difference with many paging services.
K95 probably can't be used to send pages from a modem that is connected to
a terminal server, since it has no way to set the terminal server's
communication port parameters -- speed, parity, etc -- as required by the
paging service.
Kermit 95 reportedly works fine in Windows 95 when (a) used in conjunction
with ASAP screen-reader software from Microtalk, and (b) all of the numeric
keypad keys are mapped to \Kignore. Read the K95 FAQ for details.
Beginning in version 1.1.16, K95 should work with all speech devices and
Braille devices that work with other Windows or OS/2 console applications.
Use SET TERMINAL SCREEN-UPDATE SMOOTH and SET TERMINAL SCREEN-OPTIMIZATION ON.
The Telnet server in recent versions of OS/2 (probably starting with Warp 4)
evidently does character-set translation from CP850 to Latin1, which can
interfere with the K95 Telnet client's screen formatting, especially if you
are trying to use the ANSI terminal type. Cure: Start OS/2 telnetd with the
"-cp none" parameter. Then "set term type ansi" in Kermit 95 should work as
expected. But note that function keys are not supported since they are not
part of ANSI emulation, and in any case there is no provision for them in IBM
telnetd. Ditto for Alt keys, the gray keypads, etc.
(Not a bug) In Netscape 4.0:
(Replace "c:\k95" by the path of your actual Kermit 95 directory.)
Also: The K95REGTL program was made available on October 2, 1997, for download
from the Kermit web/ftp site, and is included with K95 1.1.16. This program
can be used to register Kermit 95 as the default Telnet application to be used
by any browser at all: ftp://kermit.columbia.edu/kermit/utils/READ.ME.
In Windows 95 and Windows 98 (but not Windows NT or OS/2), if the TMP or TEMP
environment variable is defined to be the root directory of a disk,
unpredictable behavior can ensue, including crashes ("This program has
performed an illegal operation and will be shut down"), not only in Kermit 95,
but in other applications as well. This is worked around in K95 1.1.16 by
checking for this in advance and posting a warning, but the problem itself
occurs in the external DLLs that are used by K95 and other applications, not
in K95 itself. Diagnosis: Many libraries (including the TAPI and Kerberos IV
DLLs) use the TEMP or TMP definition to construct names for temporary files,
by prepending the TEMP or TMP definition to the file name, with a backslash
(\) in between. If (say) TMP is set to "D:\", and the temporary file is to be
named "tmp", this results in "D:\\tmp". Windows 95/98 mistakes this for a UNC
(which is a bug, because UNCs do not start with disk letters) and tries to
find a network node called "tmp". This can result in long delays or worse,
and at the very least, failure to create the temporary file, with results that
depend on each application or library. The bug was reported to Microsoft,
which declined to address it.
Prior to version 1.1.16, and K95 was in server mode, there was no obvious
way to send a GET command to it for a file whose name contained spaces.
Now the server allows such names to be enclosed in curly braces or
doublequotes. See
Section 5.4 of the C-Kermit 7.0
Update Notes.
You have connected to a host that is using even (or mark, or odd) parity,
but Kermit's terminal bytesize is 8 and its parity setting is NONE. This
usually happens only on serial connections direct to serial ports on UNIX
hosts. Careful, you get a terminal bytesize of 8 by default when using
VT220 or 320, or any kind of ANSI, emulation. Cure: SET TERMINAL BYTESIZE 7,
or SET PARITY to match what is used by the host.
This message is issued when a RUN command is given at the prompt, to give you
time to read the screen before returning to K95's command screen. The prompt
is not given when the RUN command is issued from a macro or a command file.
Thus you can avoid it by defining a macro that does what run does:
define xrun run \%*
and then using the macro rather than the RUN command itself. Note: \%* means
"all arguments".
Cursor positioning on double-width lines could be incorrect. Bug, fixed
in 1.1.16.
PgUp was mapped to ESC V instead of ESC v (lowercase).
Fixed in 1.1.16.
On TCP/IP connections, the SET HOST * listening port was not freed until
K95 termination. In 1.1.16 it is closed and freed when a SET HOST, SET PORT,
or CLOSE CONNECTION command is issued. And of course, also on EXIT.
If an escape sequence was received from the host that instructed K95 to change
the height/width of the Terminal screen while it was processing an INPUT
command, the physical height/width of the Command screen would also be changed
although its virtual height/width would not be. Fixed in 1.1.16.
Directory specifications such as \\node\directory were not handled correctly.
Fixed in 1.1.16.
Bug, fixed in 1.1.16.
There is no single standard interface for invoking command shells. But K95
must invoke your command shell when you give a RUN, REDIRECT, or PIPE command
since (a) K95 has no way of knowing whether the text argument of your RUN,
REDIRECT, or PIPE command is a built-in shell command, an external program, a
shell script, a batch file, or what, and (b) if it is an external program, it
is up to your shell to find it. As of 1.1.16, K95 has built-in knowledge of
the calling conventions of various alternative shells, including 4DOS, 4NT,
the Hamilton C-Shell, the Korn shell, the MKS Shell, ASH, and BASH.
Bug, fixed in 1.1.16.
Bug, fixed in 1.1.16.
Bug, fixed in 1.1.16.
SET PRINTER /BIDI was not putting all of the configuration options on the same
line. Bug, fixed in 1.1.16.
The OUTPUT and INPUT commands would fail silently under certain conditions,
e.g. if the connection was lost. As of 1.1.16 a proper error message is
issued.
SET KEY (in its multiline format) and SET MODEM would react badly to editing
via Backspace, Ctrl-W, etc. Bug, fixed in 1.1.16.
If the server sent back a "short-form reply" to a REMOTE command, and your
REMOTE command had a redirection indicator, it was ignored. Bug, fixed in
1.1.16.
When K95 1.1.15 or earlier was used for terminal emulation over a Winframe
ICA connection into a Citrix Winframe, performance was very poor. Diagnosis:
Winframe performs optimization on GUI windows but not Console windows. K95
1.1.16 overcomes this problem by doing screen optimization itself. The same
is probably true for Microsoft Hydra (the Windows Terminal Server) and its
various third-party clients ("Windows terminals" from Boundless, Network
Computing Devices, Neoware Systems, Tektronix, and Wyse). This might also
help with slowness that was reported under Windows NT with certain video
drivers.
The Windows Console driver supports only 16 colors, whereas the Wyse 370
supports 64. Prior to version 1.1.16, K95 ignored Wyse 370 coloration
directives. Beginning in 1.1.16, it attempts to map them into the 16
available colors. K95 will handle Wyse 370 colors better after conversion
to full GUI is complete.
Chinese users have reported success viewing ideographic text on the screen
when using AT386 and similar emulations (which automatically choose 8-bit
characters and no translation), they are not able to enter Chinese (or Korean,
or Japanese, etc) characters on the keyboard. This is because Microsoft chose
to disable CJK Input Method Editors (IME) in Windows 95 Console windows. See
Microsoft Knowledge Base Article Q156793: "All Far East versions of Windows
NT 4.0 support IME in a command prompt. Only the Japanese version of Windows
NT 3.51 supports IME in a command prompt. None of the Far East versions of
Windows 95 support IME in a command prompt." When an IME is not supported,
Microsoft recommends that you "enter the Chinese, Japanese, or Korean
characters into Notepad or WordPad and then paste them into the command
prompt".
It has been reported on systems with Logitech MouseWare Control Center 7.10
and driver 7.00, the mouse stops working when a screen saver is activated.
This has nothing to do with Kermit 95.
The VT 220 or 320 emulator, when in reverse screen mode (DECSCNM), would
apply the Bold attribute to the foreground rather than to the background of
the cell. Fixed in 1.1.16. Also, there was no equivalent to the VT220/320
"screen mode" setup selection (even though there was a Kverb, \Kflipscreen,
for this); this was added as SET TERMINAL SCREEN-MODE { NORMAL, REVERSE } in
version 1.1.16.
As of K95 1.1.14, the OS/2 version of the Dialer uses Work Place Shell
features that are compatible only with with OS/2 3.0 and higher.
K2.EXE itself can be used with OS/2 2.0 and later.
In Windows 95, if K95 has a transaction, session, or other log open, and is
killed from the task list, or crashes, or Windows itself crashes, the log file
might be truncated, empty, or nonexistent. This is how Windows 95 works.
This does not happen in Windows NT. To work around this Windows 95 problem,
close and reopen essential logs periodically, to force commitment to disk,
using APPEND mode in subsequent opens. Windows NT (or UNIX, etc) is
recommended for critical applications.
Code Page 856, used in Bulgaria, was evidently made up by a Bulgarian company
without checking first whether there already was a CP856. And indeed there
is: IBM Code Page 856 is for Hebrew. Furthermore, the code table for CP856,
from which K95 support was added, is evidently only one of several variations.
Testing has shown that, depending on the terminal emulation, Cyrillic letters
are OK but box-drawing characters are not, or vice versa. The Bulgarian
company has noted that this is to be expected. "CP856" is now referred to
in K95 and C-Kermit as "bulgaria-pc".
(Fixed.) The fixed description is
in the SET TERMINAL Command List.
The new optimized screen updating method might not be an improvement on every
PC. Evidently results can vary with the video adapter and driver, etc. If
you prefer the screen-updating method of K95 1.1.15, you can restore it by
putting the following command:
set terminal screen-optimization off
in your K95CUSTOM.INI file. At least one person has described a long
delay in cursor position updating which is corrected by turning off the
additional screen-optimizing calculations.
As noted elsewhere, spaces in file and directory names are a Very Bad Idea.
Thus we do not recommend installing K95 in directories such as:
C:\Program Files
One consequence of doing this is that the Kermit 95 TELNET.EXE and
RLOGIN.EXE stubs don't work when invoked from outside the Kermit 95
directory (e.g. from the Start menu). Workaround: Use the "FAT" (8.3) name
for the path when starting RLOGIN.EXE or TELNET.EXE. Or
move Kermit 95 to a directory whose name has no spaces. RLOGIN.EXE
and TELNET.EXE have been reprogrammed in version 1.1.17 to cope with
this situation.
Scroll Up and Scroll Down (SU/SD escape sequences) don't work right.
Fixed in version 1.1.17.
K95 will crash during an XMODEM transfer over a serial connection if
carrier detect drops during the file transfer. Fixed in 1.1.17.
CRC block checking must be specified by the receiver. K95 currently
provides no way to specify this. Fixed in 1.1.17.
Two bugs were accidentally assigned to number 524:
524.1 QANSI emulation bugs
The default character sets were wrong. They should be:
G0 - ASCII
G1 - DecSpecial
G2 - Latin1
G3 - DecSpecial
K95 did not interpret the character 'U' as an indicator for CP437
in the ISO 2022 escape sequences. This is a non-standard identifier.
K95 did not properly implement CSI 10/11/12 m. The previous implementation
would work since the default characters was CP437.
Cursor Horizontal Absolute was not implemented.
All fixed in 1.1.17.
524.2 Cursor position on Double Width line vs. Horizontal Scrolling
The cursor when positioned on a double-width line will be displayed
on the wrong column if K95's internal horizontal scrolling is used.
Fixed in 1.1.17
If the TMP or TEMP variables are defined but point to a non-existent
directory or a file, the dialer will crash at startup. Fixed in 1.1.17.
If no connection is in progress instatus is set to 3 (Internal Error)
when it should be set to 4 (I/O error). Fixed in 1.1.17.
The command is supposed to accept ENABLED or DISABLED with aliases
of ON or OFF. The command only accepts the aliases in 1.1.16. Fixed in
1.1.17.
OS/2 3.0 does not allow more than 32K of data to be read from the
TCP/IP socket even if more than 32K is waiting. This is due to
the use of signed ints in the TCP/IP stack instead of unsigned ints.
Version 1.1.17 restricts reads to 32K as in previous releases.
This bug does not affect Warp 4.0.
The \v(osname), \v(osrelease), and \v(osversion) variables are not
properly set on OS/2 in version 1.1.16. Fixed in 1.1.17.
When accessing a pooled IP host, each attempt is supposed to access a random
host in the pool. If Win95/98 is not configured for a WINS server it is
possible that the internal WINS cache may prevent PING, K95, and possibly
other applications from being able to randomly access a pooled host and
resolve the real name of the host being connected to. This results in
authentication failures with Kerberos 4 and Kerberos 5. Microsoft will not
be fixing this problem; they have designated it a feature.
If a machine has multiple adapters installed (physical or virtual) and
Kerberos is unable to tell which one to use, it might choose the wrong one.
If so, the IP address associated with the Kerberos ticket does not match the
IP address used to connect to the host. This results in an "invalid network
address" error and authentication failure.
Workaround: Disable or remove adapters that are not in use if possible.
This bug is also present in Microsoft Proxy Server.
FAILURE should only be reported if the PAUSE command was interrupted by the
user pressing a key. Fixed in 1.1.17.
The inability of Kermit 95 to display the Euro character is caused by two
problems:
- K95 1.1.16 and earlier did notsupport any character-sets (or code pages)
that contain the Euro character. This has been addressed in 1.1.17 by the
addition of ISO 8859-15 (IBM Code Page 923) and IBM Code Page 858. ISO
8859-15 is expected to become the new default character-set for Europe after
Jan 1, 1999. CP858 is CP850 modified to replace the dotless i with the Euro
symbol. Version 1.1.17 also supports the expanded versions of Windows Code
Pages 1250 to 1257, in which the Euro character was added in a formerly vacant
position.
- The operating system does not support the Euro symbol. Microsoft is
supposed to be adding the Euro symbol to NT 4.0 in Service Pack 4 due out
Summer 1998. IBM has replaced CP850 with CP858 in OS/2 Warp 3.0 FP35 or
higher, or OS/2 Warp 4.0 FP6 or higher. Windows 98 supports the Euro
for GUI applications but not for Console applications.
Microsoft has posted the Win95 Euro update (without Console window
support) at:
http://www.microsoft.com/windows/downloads/contents/Updates/W95EuroPatch/default.asp
Research has shown that the Monotype.com true type font available from
Microsoft's http://www.microsoft.com/typography/fontpack/default.htm
(which includes the Euro character) is available for use in console
windows on Win NT immediately after installation. This font is thinner
than Lucida Console and is therefore better suited to those sites that
need 132 columns.
The font can also be used in Win95 console Windows if the Registry
is edited and if the key
HKLM\SOFTWARE\Microsoft\Windows\CurrentVersion\MS-DOS Emulation\
is edited to read
"Font"="Monotype.com"
Charles Bigelow of Bigelow and Holmes's has also pointed out that
Microsoft could receive and distribute the latest version of
Lucida Console which contains not only the Euro but also the missing
Hebrew and Arabic glyphs for free whenever they want to from
Monotype, the company that distributes the True Type versions of
B&=38;H's typefaces.
As of Kermit 95 1.1.17, if the operating system provides support for the
Euro, then Kermit 95 is able to display it.
K95 attempts to test the connection with a function that is not implemented
in the device driver. A workaround is provided in 1.1.17.
The function does not stop splitting words at the end of the string.
Fixed in 1.1.17.
When switching from the Terminal screen to the command screen, the cursor
might disappear. This is because the Windows API for restoring the cursor
fails to work -- i.e. another Microsoft bug. Experimentation shows that
reCONNECTing and escaping back tends to revive the Command-screen cursor.
If a key is defined with either SET KEY or SET TERM KEY with a value
ending in '\' such as:
<ESC>_1<ESC>\
the '\' will not be sent. Now the way this really needs to be written
since '\' is a quote character and kverbs and variables may be embedded
into the string is
SET KEY \nnnn \27_1\27\\
but because the string gets parsed twice, once and the command line
and once before it is sent
SET KEY \nnnn \27_1\27\\\\
but this results in
<ESC>_1<ESC>\\
being sent to the host. The following workaround may be used if
you do not need the definition to be general (for all terminal types)
and the rest of the string does not contain kverbs or variables.
SET TERM KEY /LITERAL \nnnn \27_1\27\\
Fixed in 1.1.17.
The LOOKUP command does not set STATUS to FAILURE if the specified name
is either not found or ambiguous. Fixed in 1.1.17.
\frindex() would fail to match the first instance of a string if a
starting index is specified.
\frindex(r,rr,1)
returns 1 when it should return 2. Whereas
\frindex(r,rr)
returns 2 as it should. Fixed in 1.1.17.
When Unicode support is active, all extended characters (those within
the range 128-255) are displayed in CP437 regardless of the settings
of SET TERMINAL CODE-PAGE and SET TERMINAL LOCAL-CHARACTER-SET.
Fixed in 1.1.17.
It should be noted that the default code page for console windows does
not match for GUI windows. For instance, in the U.S., the default GUI
code page is 1252 while the default console code page is 437. This
affects the characters that may be produced by the keyboard and
delivered to K95. It is also important to realize that while K95 uses
Unicode as an intermediary step in translating all characters and
stores all display characters in Unicode, Unicode is only used for
display output and is not in any way used for keyboard input.
Therefore, it is very important that both SET TERMINAL CODE-PAGE and
SET TERMINAL LOCAL-CHARACTER-SET be assigned to an equivalent value.
In NT, dead keys are delivered to the application even if they are
not generating a character. For instance, ~a is delivered as two
key events. The first without a character value and the second with
a character value. Since the first comes without a value K95 thinks
it is similar to a function or gray key and when it finds no macro
or kverb assigned to it, K95 beeps. Fixed in 1.1.17.
The Visible bell was broken in 1.1.16 when implementing the screen
diff optimization. Fixed in 1.1.17.
The SET TELNET ENCRYPTION {...} command when used to set anything
other than the TYPE of encryption to be used does not work in 1.1.16.
Fixed in 1.1.17.
The SHOW COMMAND Max Command Length (10238 chars) is supposed to be
the max length of macros. However, it turns out the lower levels of
the command parser restrict this size to 4071 characters. Fixed in 1.1.17.
The syntax of the END command allows for an optional numeric value.
This value is used to set the status of the IF SUCCESS and IF FAILURE
tests. A value of '0' is SUCCESS and non-zero is FAILURE. However,
there is no way for a script to perform a switch and the actual
numeric end code that is specified which makes error handling
difficult. In 1.1.17, the END value is available in \v(return).
When the SET DIAL LC-AREA-CODES command is used to specify area codes
that should be dialed as local instead of long distance, Kermit
would forget to dial the area code. Fixed in 1.1.17.
With K95 in SERVER mode, if GET is used to transfer a file which matches
a TEXT file pattern it will be sent as TEXT regardless of the client's
file transfer setting. However, the REGET command will transfer the
file as BINARY which may corrupt the file. Fixed in 1.1.17.
If a REMOTE command fails due to an error that results in either an
E-packet being sent or received (ie, "Too many retries") it is possible
that the STATUS flag will be set to SUCCESS even though the REMOTE
command failed. Fixed in 1.1.17.
A simple script such as the following will exhibit screen update problems
for \%i 1 10 1 {
echo This is line \%i
}
run sleep 5
wait 15
This is because the Screen Update thread is not given the opportunity
to update the screen with the output from the ECHO command before the
RUN is executed. When the RUN command completes the WAIT command (which
generates no output) delays the screen update until after the WAIT
completes. Fixed in 1.1.17.
The sequence:
assign line {word1 word2 word3}
echo \fword(\m(line), 3)
would echo a blank line instead of word3. Fixed in 1.1.17.
When using partial dialing, the prefixing and suffixing rules apply
to each PDIAL and DIAL command instead of prefixing only the first.
Fixed in 1.1.17.
The PRODUCT macro, added in 1.1.16, was always given an extra parameter with
the value '1'. If the command <CSI> < 1 ; 2 ; 3 ~ is sent to Kermit 95, the
product macro will be called with four parameters and \%4 will equal '1'.
Fixed in 1.1.17.
When installing in Windows from the CDROM, the Windows "Autorun"
starts the SETUP.EXE program automatically. When the SETUP program
exits, the Console window might not close automatically. No harm done;
just click the close button [x].
It has been observed that, at least on some systems, Netscape 4.0 discards
the "/" characters from links passed to it in a DDE message, as when you
Ctrl-click on a URL in K95. This seems to happen only when the browser
was already started. Workaround: If SHOW BROWSER reveals a specific browser,
type "set browser" (with no operand) to remove it. Then all your system
defaults are used, this seems to clear up the problem (in which case,
add "set browser" to your K95CUSTOM.INI file) (K2CUSTOM.INI in OS/2).
Numerous text files were removed in 1.1.17, on purpose. All the information
in these files was incorporated into the new Second Edition Kermit 95 manual,
which is now in your Kermit 95 DOCS\MANUAL directory, and accessible
from the Dialer's Help menu. The files affected are:
DOCS\CYRILLIC.TXT Incorporated into HTML manual
DOCS\HEBREW.TXT Incorporated into HTML manual
DOCS\DIALING.TXT Incorporated into Using C-Kermit, 2nd Edition
DOCS\MODEMS.TXT Incorporated into Using C-Kermit, 2nd Edition
DOCS\TERMINAL.TXT Incorporated into HTML manual
DOCS\HOSTMODE.TXT Incorporated into HTML manual
DOCS\OS2K95.TXT Incorporated into HTML manual
DOCS\UPDATES.TXT Incorporated into HTML manual
KEYMAPS\KVERBS.TXT Incorporated into HTML manual
Note: In Windows, those who install 1.1.17 (or later) new, as opposed to
patching up to it, will also get a desktop shortcut to the manual. Those who
patch up to 1.1.17 (or later) can make a shortcut as follows:
My Computer .. k95 disk .. k95 directory .. Docs .. Manual
Right-click on kermit95.htm, choose Create Shortcut, and drag the shortcut to
the desired place.
Or K95REGTL.EXE may be executed manually in order to re-create the
Documentation Shortcuts.
The new DATE command does not return to the command prompt. Instead
the date string is printed each time you press the Enter key. Use Ctrl-C
To get out of this situation. Fixed in 1.1.18.
The CPS display in XYZ-modem transfers was broken when floating point
timer support was added in 1.1.16. Fixed in 1.1.18.
It has been reported that this setting might not work when carrier drops
during a file transfer on a modem connection. Still under investigation.
If SET FILE COLLISION RENAME is active and the incoming file requires
renaming, then after the transfer of the received file is complete K95 can
crash (or produce an unable to free memory error). Workaround: don't use
the RENAME collision option. Fixed in 1.1.18.
When using TAPI MODEM-DIALING and the number is BUSY, the next dial attempt
will fail with an "invalid call state" message. Fixed in 1.1.18.
A bug in NT's implementation of the Win32 ToAscii() function causes the
program stack to become corrupted. This results in K95.EXE's
immediate termination. A workaround is in 1.1.19.
Once a PDIAL command is issued in 1.1.17, all calls from that point forward
will be dialed as partial. Fixed in 1.1.18.
HOST.KSC defines a FAIL macro which conflicts with the FAIL command that was
added in 1.1.17. This results in two problems:
- If the userid/password combination is invalid throughout
the retry limit the user will be let on anyway with an
invalid userid
- If the connection is terminated prematurely without a
proper LOGOUT the script would enter an infinite loop.
To be fixed with an updated HOST.KSC script. Contact kermit-support if
needed. Note: whenever a macro name is the same as a built-in command,
the built-in command takes precedence. To force the macro to be run, rather
than the built-in command, preface it by DO, e.g. "DO FAIL".
In Windows 95, any command used in a pipeline must NOT be a 16-bit program,
such as SORT. If it is, K95 will hang and must be killed from the task
list, and your system will need to be rebooted. This is a Microsoft Windows
95 bug: a 32-bit program when starting a 16-bit program is unable to
determine when the 16-bit program has terminated. You can work around this
problem by creating a batch file to start the 16-bit program, and then
RUNning the batch file from K95.
Note that many 16-bit programs, such as PKZIP/PKUNZIP, are available (or
replacable) by 32-bit equivalants (such as the InfoZIP utilities). The
32-bit versions give you full access to Windows 95 features such as
long filenames.
it is always better to run native 32-bit programs under Windows 95.
In 1.1.16 and 1.1.17, SET SERVER IDLE-TIMEOUT does not work if the
connection is RELIABLE. TCP/IP connections default to RELIABLE.
Workaround:
SET RELIABLE OFF
Fixed in 1.1.18.
K95 since 1.1.7 has been incorrectly applying the Origin mode to the
Next Line command. Fixed in 1.1.18.
Due to a compilation time mixup the PCTERM support was only compiled
into test versions of K95 1.1.17 and was never built into the release
version. If you need PCTERM support contact
kermit-support@columbia.edu.
Fixed in 1.1.18.
Set the NASI Global Settings to always report hardware flow control
to the application.
As documented, VT soft character-sets are not supported by Kermit 95
due to the limitations of the console environment in which it runs.
K95 tells the host that it does not support soft character-sets when
it is queried. If the host sends a soft-character-set, it will be
ignored. If the host attempts to assign the soft-character-set to
G0, G1, G2, or G3, K95 assigns US ASCII, which can be confusing.
In 1.1.18 and later, K95 will do nothing in this situation.
As documented in the Kermit 95 HTML Manual Security section, Windows 95,
Windows 98, Windows NT 3.5x, and Microsoft Proxy Server cache DNS entries
without regard to the Time To Live (TTL) value associated with a particular
entry. The result is that Kerberos authentication fails when attempting to
connect to a hostname that is really an alias for several machines that
provide a similar function. Microsoft has chosen to issue a private fix
which may be applied to Windows 95 with the Winsock 2.0 update. The problem
is documented in the Knowledge Base as article Q192425.
The following directions specify how to receive this fix:
- Ensure that you have a valid product ID for a retail copy of the product
(right-click on 'My Computer', select 'Properties', and use the number
shown beneath the 'Registered to' text). If you do not have an ID for
a retail copy of Windows 95 (if you have all OEM versions, for example)
there will be a charge of $35 for the support incident.
- Call the appropriate number for 'Standard No-Charge Support:' for
Microsoft Windows 95:
USA: (425) 635-7000
Canada: (905) 568-4494
The best time to call is 8AM-6PM Pacific time (the support escalation
staff is online during these hours)
You will first talk to a customer service representative (CSR) who
should activate your support, and transfer you to a support engineer.
Please advise the CSR that you need to obtain a specific "hotfix for
Windows 95".
- Once you get to a support engineer, advise him or her of the following:
You need the hotfix referenced by knowledge base article:
Q192425 Gethostbyaddr() Call with Winsock2 Returns Address Pool Alias.
Hotfix fulfillment is done via an internal escalation team,
which typically does a callback to validate the need for the hotfix.
This is per Microsoft's support policy, and is strictly adhered to.
The callback typically occurs within 72 business hours from the initial
call-in.
This information was provided by a Microsoft Support Engineer.
No fixes are available for NT 3.5x, Proxy Server or Win98 at this time.
Kerberos Authentication relies on clock synchronization to determine when a
Kerberos ticket is valid for use. All clock times are stored in tickets
in Greenwich Mean Time (GMT). The C Run Time Library used the TZ
environment variable to determine the local time zone so it can convert GMT
to local time. For the Eastern
Time zone of the United States the following must be placed into the
AUTOEXEC.BAT file on a Windows 95/98 system or in the System
Environment settings for Windows NT:
SET TZ=EST5EDT
for California:
SET TZ=PST8PDT
Version 1.1.18 defaults to EST5EDT if the TZ variable is not defined.
Even though 1.1.17 allows WP keyboard mode definitions to be assigned
they are never actually used. Fixed in 1.1.18.
Sometimes it is possible to get into a truly annoying predicament, in which
Caps Lock is turned on every time you switch from one Console Window (Kermit
95 or any other console application) to another. No amount of turning Caps
Lock off makes this stop happening. As noted in
Bug #224, this is a Microsoft
bug in Windows 95 (and 98?) but not NT. Here is a workaround: Switch to a GUI
window and turn off Caps Lock there. Now your console windows are back to
normal.
The second attempt to enter CONNECT mode with a trigger will result in
a memory exception. Fixed in 1.1.18.
If you are using the following:
- A Hayes High Speed modem (V.32bis or higher) with ATE1 set
- SET CARRIER-WATCH ON or AUTO
- A script using a loop containing INPUT to read one character
at a time into \v(inchar) which then either echos the character or
OUTPUTs some other data.
When the modem loses the carrier signal, it will output the string
"NO CARRIER" and then drop the carrier signal. However, because there
is data waiting to be processed the INPUT command succeeds and does
not detect the carrier loss. Since the modem is now in the command state
it will echo data sent to it (because of ATE1) causing an infinite loop.
To detect the lost connection, add a WAIT 0 CD command followed by IF
FAILURE to the loop.
The program receives an "Out of Memory" error and terminates.
Please contact kermit-support@columbia.edu for a replacement
program.
Block mode was never implemented for the Televideo terminals in
K95 1.1.17. This capability has been added to version 1.1.18.
K95 1.1.17 would improperly handle a scroll region larger than the
display size and direct cursor addressing to the status line. This is
corrected in 1.1.18.
Observed on Windows 95B during file transfer over a Telnet session to
VAX/VMS 7.1 with MultiNet 4.0B. K95 crashes when remote host crashes.
Windows crash dialog box says:
K95 caused a stack fault in module WSOCK32.DLL at 0157:7c6ca4b8.
Stack dump shows all 0's. Winsock bug.
ASCII terminal emulations such as Wyse, Televideo and DataGeneral
have a Copy to Printer mode which prints each and every character
that is sent to the terminal display. This mode on ASCII terminals
will print double characters in 1.1.17. Fixed in 1.1.18.
When a REMOTE {SET, ASSIGN, RENAME, COPY} command is processed by
CONNECT mode it leaves the user at the command prompt instead of
returning automatically to CONNECT mode as it should. Fixed in 1.1.19.
When the following conditions are true file sends may result in a K95
crash due to an "invalid access" error:
SET TAPI MODEM-DIALING OFF
SET TAPI LINE device
SET WIN95 OVERLAPPED-I/O ON
SET FLOW not specified to match that used by the modem
The problem is that when the TAPI device is allocated the transmit
buffers are not properly allocated. Fixed in 1.1.19.
(OS/2 only) When the SERVER receives a REMOTE HOST command it does not
correctly pass the command to be executed to the command shell.
A workaround. If you know you are communicating to OS/2 K95 version
1.1.17, send the command to be executed as
REMOTE HOST /c
instead of
REMOTE HOST
Fixed in 1.1.19.
PC Code Pages are not compatible with international standards for
character sets. Code Pages place printable characters in the range
allocated for the C1 control characters. The C1 control character CSI
(0x9B) is the equivalent of ESC [. When performing transparent printing
the emulator must process CSI sequences to look for CSI 4 i to determine
when printing should terminate. However, with code pages the code point
0x9B should not be treated as CSI because it is a printable character.
1.1.17 forgot to treat 0x9B as a printable character when performing
transparent printing. This is corrected in 1.1.19.
The MIT Kerberos 2.0 for Win32 includes a new Kerberos 5 1.1 compatible
library. This library is missing a function that was available in the
previous NT Alpha 2 distribution of Kerberos 5 1.0.
K95 1.1.17 has a bug that will result in all telnet authentication
being disabled if it finds Kebreros 5 libraries but cannot successfully
load all of the necessary functions.
This problem is fixed in K95 1.1.19.
K95 1.1.17 was designed to work with SRP 1.4.1 and simultaneously
be compatible with the most up to date IETF Internet Draft on Telnet
Authentication. SRP's Telnet Authentication did not adhere to the
Internet Draft. The SRP Telnet authentication handshake did not
protect against a man in the middle attack which would allow
authentication to succeed and encryption to be disabled. In the
process of fixing this security hole the SRP Telnet authentication
handshake was altered. In the process it broke compatibility with
Kermit 95 1.1.17's implementation of telnet authentication.
If you are connecting to a host that uses SRP 1.5.0 Telnetd you
will not be able to authenticate using K95 1.1.17. K95 1.1.19
is compatible with SRP 1.5.0.
Kermit 95 1.1.17 included a new FAIL command, which conflicted with the
FAIL macro that was defined and used in HOST.KSC. Shortly after
K95 1.1.17 was released, a corrected version of the host-mode script was
put on the Kermit FTP site and downloaded by most people who used host mode.
Unfortunately, this means that HOST.KSC will not be updated in the
1.1.19 patch. If you downloaded the fixed HOST.KSC file for
K95 1.1.17 and then installed the 1.1.19 patch, you'll need to download the
1.1.19 version of HOST.KSC after installing the patch:
ftp://kermit.columbia.edu/kermit/k95/host.ksc
(Not a bug.) If you attempt to run old host-mode scripts with a newer
version of Kermit 95, they are likely not to work. For example, the way
Telnet negotiations are handled changed in K95 1.1.19, and the
HOST.KSC and HOSTTCP.KSC scripts were modified accordingly.
The new scripts are likely to fail with syntax errors under older K95
versions, and old scripts are likely to hang under the new K95 version. The
host-mode scripts are updated automatically when you install an upgrade patch
UNLESS YOU CHANGED THEM. Modified files are not, and can not be, patched. In
this case, you can download the current scripts from the Kermit FTP site (previous item).
A new feature in K95 1.1.19 and C-Kermit 7.0 is the ability to include
parameters for command files on the TAKE command line, documented
in Section 7.16 of the C-Kermit 7.0 Update
Notes. As noted there, any arguments you give to a TAKE command are
persistent. This can produce surprising results with subsequent TAKE
commands. For example:
K-95> take foo.ksc arg1 arg2 ...
K-95> take bar.ksc
The second command file, bar.ksc, receives the same arguments as the
first. This is not a bug, exactly, but it's not good either. Future releases
of C-Kermit and K95 are likely to change this behavior, so don't depend on it.
In the meantime, you can work around any problems caused by the current
arrangement (e.g. when the second TAKE file refers to \%1,
\%2, etc, but you don't include arguments when TAKE'ing it) by
explicitly clearing the affected variables:
K-95> take foo.ksc arg1 arg2 ...
K-95> undef \%1
K-95> undef \%2
...
K-95> take bar.ksc
It never did. Console applications do not receive a Close message from
Windows 9x when you click the Close [x] box. The application is terminated
abruptly by Windows, leaving all its resources (such as the TAPI device)
in an indeterminate state, and Windows itself unstable until the next reboot.
In K95 1.1.19, a method was found to disable the Close [x] box to prevent
this from happening; the [x] box is simply grayed out and inoperative. Since
it is not possible to receive a message when the user clicks this box, there
is no way to issue a warning or give instructions on the proper way to exit
K95.
A secondary problem with the disabled Close box is that it remains disabled
even after K95 exits. You will notice this only if you start K95 from an
MS-DOS Command window. Terminate the MS-DOS Command window with an EXIT
command at the DOS prompt, or with Ctrl-Alt-Del.
K95 Bug. Workaround: Use VT220 and specify VT102 as the Telnet terminal
type if VT102 must be used. Fixed in 1.1.120.
K95 1.1.19 is stricter about enforcing Telnet protocol rules than previous
releases, and therefore might fail to make connections with faulty servers
when previous releases could do so. In most cases, the old behavior can
be restored by including the /NOWAIT switch in your TELNET or SET HOST
command, or (globally) with SET TELNET WAIT OFF. See
http://www.columbia.edu/kermit/telnet.html
for details.
After applying the 1.1.19 upgrade patch to OS/2 Kermit 95, the OS/2 Dialer
was corrupting the DIALUSR.DAT user database. This was fixed by
replacing the OS/2 patch for 1.1.19 and notifying the affecting users.
There was a typo in the LOGIN.KSC file that prevented its use when
not used in conjunction with the dialer. Fixed in 1.1.20.
Code Page character sets have printable characters in the C1 range. These
characters are misinterpreted when they are printed in the rightmost column of
VT terminal emulations. Fixed in 1.1.20.
IBM 3151 attributes can be applied as bitmasks with replacement, bitwise OR
and bitwise AND operations. Kermit was treating all attributes as
replacement. Fixed in 1.1.20.
K95 did not properly seed the PRNG. Fixed in 1.1.20 along with an upgrade to
OpenSSL 0.9.5a (Win32 only; OS/2 is still using 0.9.4).
The NRL Kerberos DLLs use a slightly different interface than the MIT DLLs
upon which they are based. 1.1.20 supports both interfaces and specifically
supports the NRL distribution.
The address field of the MAIL command was lost during parsing. Fixed in
1.1.20.
Workaround: send ../*.c. Fixed in 1.1.20
In a macro, if a variable "abc" was declared LOCAL when there was
an existing global variable "abcdef" the following side effects
were observed:
- The global variable would be deleted when the macro was executed.
- The local variable would become global upon macro termination
and would be assigned the value of the deleted global.
Fixed in 1.1.20.
Kerberos password prompts were being displayed in popups when K95 was started
with the -# 96 command-line option (use standard input/output),
even though popups do not work when K95 is instructed to use standard i/o.
Fixed in 1.1.20
The ANSWER command with a timeout value of 0 is supposed to wait forever for a
call. Instead, it was timing out after 254 seconds. Fixed in 1.1.20.
If a TAPI device was being used with TAPI Modem-Dialing set to ON, the ANSWER
command would fail to answer the call reporting "Operation Unavailable". This
is due to a change in behavior of newer TAPI service providers. Fixed in
1.1.20.
1.1.19 did not send Rlogin Window Size reports. Fixed in 1.1.20.
Kermit provides a method for automatically determining by country code whether
Pulse or Tone dialing should be used by default. K95 1.1.19 had bug resulting
in the automatic mode not being overridden by explicit SET DIAL METHOD PULSE
or TONE commands. This more often than not results in Tone dialing being used
under all circumstances in country codes:
1 31 32 33 352 353 354 358 39 44 45 46 47 49
Fixed in 1.1.20.
There is a bug in the MIT Kerberos for Microsoft Operating Systems Release 2.0
which will result in a crash of Kermit 95 if the new /NO-ADDRESSES switch is
used to retrieve tickets without embedded IP addresses. The problem (and a
fix) has been submitted to MIT for inclusion in the next release.
The previous fix to RLOGIN window size notifications (bug 605)
appears to have made things worse. Fixed in 1.1.21.
The prerelease versions of the Kerberos 5 libraries do not work with the
initial release of 1.1.20 as they are missing necessary functionality to
support the AUTH K5 INIT /NO-ADDRESSES and Hardware Pre-auth functionality.
It was not intentional that older releases should no longer be usable. A
future patch may restore this functionality with the limitations that
/NO-ADDRESSES and Hardware Preauth cannot be used.
The SET FILE DOWNLOAD-DIRECTORY command when issued without a directory name
could lead to very unexpected results. Instead, of clearing the
download-directory it would assign a directory of the null-string. How this
causes Kermit to behave is unpredictable. This was fixed in 1.1.20.
The HOSTMODE scripts for 1.1.20 might not be able to find either the k95.exe
executable or the host.cfg file depending on how it is started. Please
download:
ftp://kermit.columbia.edu/kermit/k95/hostmode.ksc
ftp://kermit.columbia.edu/kermit/k95/host.ksc
ftp://kermit.columbia.edu/kermit/k95/hostmdm.ksc
ftp://kermit.columbia.edu/kermit/k95/hostcom.ksc
ftp://kermit.columbia.edu/kermit/k95/hosttcp.ksc
and replace the copies in your k95\scripts\ directory.
Please download:
ftp://kermit.columbia.edu/kermit/k95/hostmode.bat
and replace the copy in your Kermit 95 Program directory. Fixed in 1.1.21.
If the telnet server requests the use of Forward-X, K95 negotiates it
even though it can't find an X Windows Server. This could lead to dual
mode programs such as emacs to attempt to open an X Display when a
display does not exist. To be fixed in 1.1.21.
The table of default keyboard definitions used to initialize the
strings sent by Wyse keyboard verbs is incorrect for all keys
other than F1 to F16. These values are used both for initial
key assignment and when resetting the key values under host command.
Fixed in 1.1.21.
The krb5 kermit variables report the status of the last kerberos 5
library call. After some errors occur, other kerberos 5 library
calls are made that clear the error values. Corrected in 1.1.21.
The names used in fields for file names do not support the %ENVVAR%
style environment variable expansion used in the .BAT batch files.
Not only does this notation not work in Kermit 95 but it can cause the dialer
to perform a fatal exception. Fixed in 1.1.21.
The SET PRINTER command has evolved since the introduction of Kermit 95
in October 1995. The original SET PRINTER command was extremely simple:
SET PRINTER dos-name
where dos-name could be any filename or device name that could be
opened via the C run time library fopen() call. By the time 1.1.17 was
released the SET PRINTER command had been modified to support piped commands,
windows printer queues, and numerous configuration values for timeouts,
bidirectional printing, ... The problem with 1.1.17 was that if you wanted to
change the printer timeout value you would need to re-enter the entire SET
PRINTER command line to respecify all of the other configuration values.
So in 1.1.18 and higher, the SET PRINTER command switches are sticky.
This means you can now issue a sequence of commands such as
SET PRINTER /WINDOWS-QUEUE: ; or /WINDOWS: or /WIN: or /W:
SET PRINTER /TIMEOUT:300
SET PRINTER /END-OF-JOB-STRING:\12
instead of combining them all on a single line. This is wonderful except that
there is now confusion when using a mixture of commands. The help text for
SET PRINTER states:
If you give a printer name without specifying any of these switches, then it
is assumed to be a DOS printer device or filename unless the name given (after
removing enclosing braces, if any) starts with '|', in which case it is a
pipe.
but the following sequence (in 1.1.18 through 1.1.20):
SET PRINTER /WINDOWS-QUEUE:
SET PRINTER PRN:
does not "choose the default windows printer queue" and then "choose the DOS
device PRN:". Instead this sequence is treated equivalent to
SET PRINTER /WINDOWS-QUEUE PRN:
which results in an invalid printer configuration since there is no windows
printer queue with the name "PRN:". Not that this should be allowed
either because "PRN:" is not a valid printer queue and K95 should
detect that (but it does not.)
This situation becomes a problem when there is a mixture of SET PRINTER
commands from a variety of sources: K95CUSTOM.INI file, the Dialer
Connection Script, and a keymap or login script taken by the Connection
Script. For instance, if the K95CUSTOM.INI file has a:
SET PRINTER /WINDOWS-QUEUE:
command which selects the default Windows Printer Queue and a subsequent
script contains a
SET PRINTER LPT1:
command using the old format, the result is a printer
configuration that does not work.
If printing does not work when you expect that it should, use the:
SHOW PRINTER
command to examine the current configuration. If it looks like:
[D:/kermit/src/] C-Kermit> sh print
Printer: prn:
/OUTPUT-ONLY
/WINDOWS-QUEUE
/TIMEOUT:0
/END-OF-JOB-STRING:(none)
/JOB-HEADER-FILE:(none)
where the printer name is correct but it is not a Windows Queue
and yet the /WINDOWS-QUEUE flag is set; it means that you need
to use the command form
SET PRINTER /DOS-DEVICE:name
when setting the printer. Track down the script which contains
the:
SET PRINTER name
command and update it.
Version 1.1.21 restores the proper functionality of the:
SET PRINTER name
form of the command and also confirms the validity of the name if
it is a Windows Printer Queue.
It should also be noted that the text describing the SET PRINTER
command in the K95 HTML manual is a bit confusing because it
describes the printer type switches as have the format
SET PRINTER /WINDOWS-QUEUE[:]
when it should be described as:
SET PRINTER /WINDOWS-QUEUE[:[]]
where providing the colon without a name implies the default printer
name of the given type.
The userid field on the RLOGIN command is supposed to allow the
specification of an alternate name to use on the remote host.
Rlogin sends both the local and the remote names to the host. In
all versions of K95 after 1.1.17 the names are swapped.
RLOGIN /K5 does not allow the userid to be input.
This is fixed in 1.1.21.
The -M option is supposed to allow you to assign an alternate
userid (SET LOGIN USER name) from the command line. This switch
has no effect in 1.1.18 to 1.1.20. To be fixed in 1.1.21.
Not really a bug. Spaces in path names are very hard to work with in
scripting languages that use whitespace as token separators. Windows provides
a mechanism to determine the short names for a given path. In 1.1.21 and
greater these variables always use the short names instead of the long names,
and new functions, \flongpathname() and \fshortpathname(),
are provided to convert between the two forms.
CJK versions of Windows 95/98 do not (by default) install an Input Locale
which supports any of the Latin Code Pages (CP 1250-1258). Since the Windows
95/98 Console mode does not support Unicode K95 must perform translations of
all keyboard events to one of the Latin Code Pages. Without a translation
table all keyboard events return error values and K95 is unable to process
keyboard input.
In 1.1.20 and earlier, if an Input Locale for English, German, French,
... is installed it will still appear that input cannot be processed
since the default Input Locale (CJK) does not include the Latin
alphabet. Pressing Left Alt-Shift one or more times until the Latin
language Input Locale is selected allows for proper processing of
keyboard input. 1.1.21 and later support the following behavior:
- K95 tests each installed Input Locale until it finds one that is able to
translate the letter 'a', at which time it stops and uses that Input Locale as
the default for the session.
- if no input locales are available that can translate the letter 'a',
Kermit 95 terminates with an error message stating that an Input Locale must
be installed.
- K95 performs this test only if a user presses a key. This way if K95 is
being run from a script or using stdin the test is never performed.
This is a design decision, not a bug. The QNX operating system toggles the
value of the Keypad-Plus and Keypad-Minus keys when the NumLock key is
pressed; Windows does not. Therefore, the keys can only have one of the two
assignments. K95 uses the assignments recommended by Dan Hildebrand of QNX in
his document "Terminal Control under QNX" dated Nov 14, 1991.
The default mapping is to assign the SELECT function to Keypad-Plus
and CANCEL to Keypad-Minus:
SET TERM KEY QNX /LITERAL \363 \{255}\{167} ; SELECT
SET TERM KEY QNX /LITERAL \365 \{255}\{163} ; CANCEL
and to leave the "+" and "-" strings assigned to the Shifted keys:
SET TERM KEY QNX \875 \{43}
SET TERM KEY QNX \877 \{45}
These mappings can be reversed by adding the following to the
K95CUSTOM.INI file:
SET TERM KEY QNX /LITERAL \875 \{255}\{167}
SET TERM KEY QNX /LITERAL \877 \{255}\{163}
SET TERM KEY QNX \363 \{43}
SET TERM KEY QNX \365 \{45}
The IBM 3151 is a 7-bit terminal with two character set storage areas.
Both are set to ASCII by default. Once a storage area has been set
to the Graphics character set, it cannot be restored to ASCII.
One of Kermit's strengths is its ability to handle translations between
character-sets. With the release of 1.1.19 (and 1.1.20) we altered the
program to use Unicode (UCS2) as the intermediate character set for all
translations. Unlike traditional character-sets UCS2 requires that its data
be represented in 16-bits instead of 8-bits. While this allows UCS2 to be the
perfect intermediary character set since it includes all the characters in all
other character sets the use of 16-bits means that the representation of the
character in memory becomes dependent on the byte ordering of the underlying
computer architecture: Big-Endian or Little-Endian. When the byte orders
become confused translations give bad results.
In 1.1.20 the TYPE command can be used for ascii text but cannot
be used to display text in alternative character sets such as UTF-8,
UCS2, or files from other platforms.
In 1.1.20 the TRANSLATE command can be used to display files of
any type to the display, but due to byte ordering issues it cannot
be used to actually perform a translation without using a work around.
The command
translate infile incs outcs outfile
results in garbage. To get this to work in 1.1.20 you need to
temporarily alter the byte ordering
set file ucs byte-order big-endian
translate filename hp-roman8 cp1252 newfilename
set file ucs byte-order little-endian
This is fixed in 1.1.21.
If you are using K95 to access the SFU v2 Telnet Server, do not use
VTNT terminal emulation. Microsoft has altered the contents of the
data structure and it is no longer compatible with the VTNT implementation
in 1.1.20. This is fixed in 1.1.21.
If you have a wheel mouse, the numbering of the buttons might not be what you
expect. For example, the left button might be button 3 and the wheel might be
button 2. Some wheel mice allow the wheel to be clicked (or double clicked).
Since the default assignment for the \Kpaste verb is Mouse Button 2
Double-Click, users of wheel mice might need to change this assignment if they
want it to apply to the left mouse button. If your Control Panel has a Test
Mouse feature, you can use it to find out what the button assignments are.
Also see Item 265 in the main K95 Bug List.
Invalid escape sequences are sent to the printer without the ESC character.
Fixed in 1.1.21.
This may be caused by an out of memory condition in the User Resource Heap.
The User Resource Heap is a limited amount of special memory (64k) into which
all applications must store their GUI objects such as Windows, Buttons, List
Boxes, etc. The Dialer's Edit Dialog in 1.1.20 contain over 300 window
objects and requires 23k of memory. If this much memory is not available, a
GPF occurs. Fixed in 1.1.21.
Windows 95/98 come with SECUR32.DLL to provide the Security Support
Provider Interface. This is a standard part of the Windows API. If this file
is missing, K95.EXE crashes upon startup. Fixed in 1.1.21.
On a small number of Windows 95 systems it has been reported that there is a
non-fatal GPF produced in KERNEL32.DLL when the Dialer is started.
This has been traced to a defect in some versions of IPHLPAPI.DLL.
One known version of the DLL with the problem is version 5.00.1952.1. You can
download a newer version of the DLL that does not suffer the problem from
ftp://kermit.columbia.edu/kermit/k95/iphlpapi.dll
Either replace the file located at \WINDOWS\SYSTEM\IPHLPAPI.DLL with
this newer version or place the file in the K95 directory.
If K95 has a printer open when it terminates, the printer is not closed. When
printing to a file, this can result in the print output being truncated.
Fixed in 1.1.21.
We've said it before and we'll say it again: Winmodems should not be relied
upon for any serious work. For testing new dialin modem pool banks at
Columbia, we have a script that repeatedly dials, logs in, uploads and
downloads a 100K ZIP file, logs out, and hangs up. When this script is used
with a Winmodem, the rate at which successful calls were disconnected before
the data transfer phase was complete was about 13% (1700 trials). The same
script run with an equivalent external (real) modem on the same PC encountered
zero (0) disconnections. Furthermore, when Winmodems are in use, they can
impact your PC's performance in ways that an external modem can not: Windows
freezes periodically; the keyboard buffer is scrambled, etc.
A bug in the SRP client library that opens SRP to attack by a fraudulent
server has been discovered. This is not a deficiency in the SRP
authentication protocol but a programming error in the library which has been
fixed as of SRP release 1.7.0. The fix for this release is incorporated
into the 1.1.21 release of K95.
The Start Menu shortcut for the Bugs file is created for the file named
bugs.htm and not k95bugs.htm as it should be.
The 1.1.19 version of UPDATES.HTM did not get updated for 1.1.20.
1.1.20 added:
- Kerberos 5 user to user authentication
- support for NRL Kerberos 5
Due to changes to the definition of Linux Console in the Linux Kernel built
into Red Hat 7.0 and later, the K95 Linux console implementation cannot be
used with 'vi' and perhaps other applications. This is corrected in 1.1.21.
It was found that a modem response message greater than or equal to 128 bytes
in length causes a buffer overflow in K95 versions released over the last
several years. This is corrected in 1.1.21.
If a SOCKS server is being used, and the SOCKS server has multiple IP
addresses, and access to the desired host is denied or otherwise unavailable,
then K95 would attempt to connect the SOCKS server's alternate IP addresses as
if they belonged to the desired host. This is fixed in 1.1.21.
If you can't start K95.EXE from within a DOS Prompt or a DOS
application it is most likely because that session is configured to prevent
it. See:
Cannot Start a Windows-Based Program from MS-DOS Prompt
http://support.microsoft.com/support/kb/articles/Q134/4/89.asp
Apparently, some versions of Windows 98 Second Edition have the:
[x] Prevent MS-DOS based programs from detecting Windows?
flag set in the _Default.PIF configuration program. The bizarre
side-effect of this is that our SETUP.EXE program which is 32-bits cannot
start sub-processes (e.g. K95PACK.EXE) because COMMAND.COM
is executed by the system() API, and COMMAND.COM is an
MS-DOS based program.
The K95 Dialer uses the TEMP or TMP variables to determine where to create the
temporary connection scripts used by K95.EXE to connect to a host.
If the path contains spaces a command line of the form:
k95.exe c:\temp path\script.ksc
is created. Since the filename is not surrounded by quotes K95.EXE
treats the command line as two parameters; one of which is illegal. This is
fixed in 1.1.21.
K95D.EXE contains a resource leak. Handles to the Process and
primary thread spawned when starting K95.EXE were never closed.
Windows eventually runs out of Handles or Process space. This prevents
additional K95.EXE sessions from being spawned until the
K95D.EXE process is shutdown and restarted.
On Windows 95/98/ME, there is a strange behavior in the CreateFile()
API when given the OPEN_ALWAYS flag. It is supposed to return
ERROR_ALREADY_EXISTS when the file exists. Unfortunately, this is not always
the case. When this bug is triggered, the file being received has
both an original and backup file created, both empty. Fixed in 1.1.21, by
no longer relying on the documented behavior.
A just-discovered bug which has been present since version 1.1.3 can result in
scripting commands such as INPUT and OUTPUT to fail if a SET SPEED command was
not executed after the SET TAPI LINE which opened the device. This is fixed
in 1.1.21.
The Dialer, when configured to use a direct serial connection, fails to output
SET STOP-BITS defaulting to 1 and also fails to output SET PARITY if the
setting is "Even_8Bytes", "Odd_8bytes", "Mark_8bytes", "Space_8bytes".
In Windows XP, the default location is not fully specified to the application.
K95.EXE 1.1.20 and earlier do not check for an incomplete location
and crash on startup if one is found. This bug is fixed in 1.1.21.
Workaround: Use the "Phone and Modem Options" control panel item to create a
location.
When using certain fonts in Windows NT, 2000, or XP in the Console version of
Kermit 95, character bits and pieces can be left behind when a cell is
cleared, especially if the cell was highlighted and was adjacent to a block
cursor (as often happens with Unix more, Lynx, or various news
readers), or if it has been selected with the mouse. This has been observed
consistently, for example, with the Lucida Console font in Windows XP. It's a
Windows problem. Workaround: to clear the ghost fragments, type Alt-x twice
in a row. To avoid the problem, pick a different font. This won't happen in
the GUI version.
TIMESYNC.EXE uses RFC868 Time protocol (1983). Increasingly,
however, Internet time servers are shifting to Network Time Protocol (NTP,
the topic of about twelve RFCs from 1985 to the present: see http://www.eecis.udel.edu/~ntp/index.html,
which is a web site maintained by the authors of NTP for recommendations and
software for synchronizing their clocks. Windows XP supports NTP as part of
its "Date and Time Properties" control panel. See the "Internet Time" for the
ability to "Automatically synchronize with an Internet Time Server" or "Update
Now". Custom servers may be specified.
The following:
define Bob {
for \%i 1 1 1 {
output \{13}
}
}
do Bob
gets an "unbalanced braces" error (because of the quoted backslash).
Workaround: avoid \{nnn} notation for characters inside brace-enclosed
loops. Fixed in version 2.0.
For some people, under certain as-yet unknown circumstances, the terminal
screen cursor can disappear. This is not a new problem, but we have never
been able to identify a scenario to reproduce it. Workaround: two Alt-X's in
a row to exit and reenter the terminal emulator, or do anything that would
resize the terminal window. If you can give us a sequence of events that
reliably reproduces this problem (or if we find one ourselves), we can fix it.
The termininal-screen screen status looks like this if you type the
CONNECT-mode escape character
ESCAPE Command: Prompt: c Help: h
It should say "Command Prompt: c" instead of "Command: Prompt: c".
If, after typing the CONNECT-mode escape character, you type a second copy of
the escape character, the terminal-screen status line is not restored. Fixed
in version 2.0 Workaround: Use Alt-R to reset the emulator.
"help remote host" says synonym is RHELP instead of RHOST.
Fixed in version 2.0.
Fixed in 2.0.
Apparently this has been broken through the entire 1.1.20 release period
and no one ever complained. It's funny what a single missing colon can do
in a protocol message. Fixed in version 2.0.
There should not be a browse button for the TLS Ciphers field.
The TLS Settings page cannot be used with FTP connections even
though K95 supports SSL/TLS secured FTP sessions.
The paths generated with the Browse buttons should use forward
slashes instead of the traditional backslashes found on Windows.
These are fixed version 2.0.
Both commands incorrectly generate the new k95.exe command line.
Fixed in version 2.0.
Dialing does not work consistently when carrier detect watching
is enabled. Fixed in version 2.0.
When SET TERMINAL KBD-FOLLOWS-GL/GR is OFF
Russian and Hebrew keyboard modes don't work.
Off is the default mode. Turn it ON to work around the bug. This is
fixed in version 2.0.
Only the first local and remote forwarded port is listed in the
display. This is fixed in version 2.0.
The functionality works there is simply no visible feedback on the
status line. Fixed in version 2.0.
Internally ASK is set to the wrong numeric value and due to the
order of value testing invalid values are treated as OFF. This means
that invalid host keys will be accepted without warning. This is a
security hole in K95. Fixed in version 2.0. To work around,
use ON rather than ASK.
This is a bug introduced in 1.1.21 with no work-around. Fixed in version 2.0.
SHOW KEY ... { RUSSIAN, HEBREW } report success but the wrong mapping
table was being referenced. There is no work around. Fixed in version 2.0.
If you launch a K95G connection from the Dialer, and the Dialer entry changes
the terminal dimensions, and then if you move the K95G window (and only the
first time you move it), the Terminal screen is cleared (but not lost; the
cleared material is still available in the scrollback buffer). Cause and
cure unknown.
It is conceivable that the presence of Everson Mono Terminal on your PC might
affect other applications, such as Netscape, that search for appropriate fonts
to display text in various scripts. In such cases, if the result is
undesirable, you might need to reconfigure the affected applications, or
uninstall Everson Mono Terminal (in the Window Control Panel Fonts dialog).
(Not a bug) If you disable screen updates while moving windows (e.g. in
Windows XP Desktop → Properties → Appearrance → Show window
contents while dragging), then in K95G when Actions → Resize Mode
→ Changes Dimensions is selected the popup that shows the changing
dimensions is not updated, and with Actions → Resize Mode →
Scales Font, you can't watch the font change size.
K95G and the Dialer have Help→Manual dialogs, which bring up the Kermit
95 manual in your Web browser. This works fine for most users, but several
users have reported various failures or problems with interaction between K95
or the Dialer and browser; these have to do with the particular browser and
version, the particular OS and version, and K95's SET BROWSER value, if any.
K95G and the Dialer are using totally standard APIs for invoking the browser.
In K95 2.0, SET TERMINAL ATTRIBUTES UNDERLINE OFF does not disable underlining;
SET TERMINAL ATTTRIBUTE BLINKING OFF does not disable blinking. Fixed in
2.0.1.
In K95G 2.0, when the Window is Maximized and the Resize Mode is "Scales
Font", if the host changes the line width, the font did not resize properly.
Fixed in 2.0.1.
There was no SET GUI WINDOW RESIZE-MODE command in K95G 2.0 to match the
Actions→Resize Mode menu item. Fixed in 2.0.1.
COM-port connections (but not TAPI ones) could become disconnected upon
escaping back to the Command screen and then re-entering the terminal screen.
Fixed in 2.0.1.
There is no method of disabling the functionality of the Toolbar.
Copying the registration info from your old K95 version to 2.0 could fail
if your old registration did not include a company name. Fixed in 2.0.1.
[ Top ] [ Index ] [ K95 Home ] [ Kermit Home ]
Kermit 95 Bug List / Columbia University / kermit@columbia.edu /
7 June 2002