CURRENT KNOWN BUGS WITH AMIGA KERMIT No DIAL or SCRIPT commands yet. The perror() command used under 3.03 does not give helpful error messages, only error numbers, if available. Depending on unknown factors of AmigaDOS buffering, it is possible that wildcard expansion nested inside several levels of directories will cause significant disk thrashing. I have not seen this yet, and have avoided complicating the expansion code by modifying it to not thrash. When the Amiga's serial output is looped back to its serial input, (through a modem, say) and the serial device is set to a high baud rate (>100K), when a break is sent, the Amiga will hang up. The setup and cleanup of Kermit's window does not currently support Workbench, and the default window created by Lattice's _main is not eliminated. This is easily fixed, but there are other problems with Workbench invocation. See below. When Kermit is invoked from a CLI and spawns a CLI subprocess with '!', after ENDCLI is used to terminate the subprocess, the ENDCLI command line is written to the console window of the original CLI process, without a trailing newline. This seems to be an AmigaDOS quirk, and is primarily an annoyance. When Kermit is invoked from Workbench, however, the Amiga crashes immediately after the ENDCLI command (although the problem may be unrelated). The current directory name is constructed from the previous current directory name and a new relative name. This is similar to the way the CD command operates, but it is subject to some peculiarities. For instance, 'CWD C:' followed by 'CWD /' results in the current directory being the the parent of C: and a directory name of 'C:/'. It is possible to reconstruct the name using the Parent() AmigaDOS function. (This would also allow Kermit under the Workbench to reconstruct the current directory name.) However, the V1.1 RAM: handler has bugs in the Parent() function, and so would probably not function with this method. Under 3.04 Lattice C, the Lattice functions are used for chdir() and getcwd(), since they seem to use Parent() reconstruction. (Lattice 3.04 C in general assumes use of the fixed V1.2 RAM: handler.) There should be some form of output control and input for one-liners invoked with the '!' command. Due to a bug in all C-Kermit's, the transaction timer is not reset before some remote commands are executed on some other versions of Kermit. The reported transaction time will be the elapsed time since the timer was last reset, or the base (zero) time of the system, midnight January 1, 1978 on the Amiga. This can be demonstrated by starting C-Kermit, requesting REMOTE HELP from a VMS Kermit-32 version 3.2.077 server before any file transfer, and then requesting STAT. An audible bell is needed. The use of 'SET MODEM-TYPE type' to select modem control usage will have to change when the DIAL command is supported. The 23 by 77 window in connect mode is incompatible with the assumptions of many editors. Emulation of a real terminal like a "not quite" ANSI VT100 would be nice. Kermit should support AmigaDOS wildcarding, especially for SEND's. It can be argued that Unix wildcarding is often appropriate for remote GET's from the Amiga server, but AmigaDOS wildcarding should be optionally available. 5-Sep-86 01:52:03-EDT,1057;000000000001 Return-Path: Received: from CUVMA.COLUMBIA.EDU by CU20B.COLUMBIA.EDU with TCP; Fri 5 Sep 86 01:52:00-EDT Received: from TUCC(MAILER) by CUVMA (Mailer X1.23b) id 4140; Fri, 05 Sep 86 01:52:08 EDT Date: Fri, 05 Sep 86 01:50 EDT To: kermit@cu20b,fdccu@cuvma Reply-To:ccphil@TUCC From: CCPHIL@TUCC I have found the following kermit initialization file useful for the Amiga, when you want to emulate a standard ANSI terminal (like a VT100). I have used this with a PCI protocol converter and also a VAX VMS system. The PCI works well, but the VAX has problems in EDT when lines need to be scrolled up from the bottom of the screen. Note that the Amiga Kermit will configure itself based on the settings in the Preferances screen for the serial line. Insert the following lines into a file s:.kermrc : % % Set up to 80 columns by 24 rows, position to (0,0), & clear screen % echo \\033[25t\\033[80u\\033[0x\\033[0y\\014 % echo Kermit is initialized and 80x24 ANSI screen! ------------------------------ Date: Sat, 25 Apr 1987 00:50 CST From: Ian Horstmeier Subject: Bug in Amiga C-Kermit Keywods: C-Kermit I recently received this from a friend of mine regarding C-kermit for the Amiga. Perhaps this information will be useful. Subject: Re: Using VT100/KERMIT and IBM systems Summary: The fix for C-Kermit The reason that C-kermit on the Amiga doesn't work with the IBM, is because the parity is not set! I got my version of C-Kermit from kermserv@cuvma on BITNET, and it still contains this bug. In the CKICON.C module, look in the connect mode function. In about the middle of it, there is a call to the output a character function ttocq(c). This needs to be changed to ttocq(dopar(c)). There! It now works with IBM and sets parity correctly. According to the comments in the code, Kermit does its own parity checking and the serial.device is always in 8 bit no parity. I always use kermit on the mainframe and the vax. One thing you will notice is that the standard amiga keymap does not generate codes to be compatible with anything! I am in the process of writing a keymap module for amiga c-kermit to make it look like a vt100. Good luck! Walter Reed UUCP : ihnp4!umn-cs!ndsuvax!ncreed Bitnet: ncreed@ndsuvax OR NU105451@NDSUVM1 ------------------------------