Date: Thu 6 Aug 87 12:19:49-EDT From: Frank da Cruz Subject: Announcing C-Kermit 4E(066), a Test Release To: Info-Kermit@CU20B.COLUMBIA.EDU This is to announce an experimental new release of C-Kermit for Unix, VAX/VMS, the Apple Macintosh, Apollo Aegis, the Commodore Amiga, and Data General AOS, AOS/VS, and possibly other Data General operating systems like RDOS. I've tested this version on various Unix systems (Ultrix 1.2 & 2.0 on various VAXes, AT&T System V on a 3B20, and 2.9BSD on a Pro-380), but not on anything else. Since I'm about to leave on vacation for several weeks, I'd appreciate it very much if during that time people could try it out on all the other systems it hasn't been tested on, including Macintosh, Apollo, Amiga, VMS, Data General, and many Unix variants (Xenix, Venix, Zeus, etc, especially in local or dialout mode), and report back to Info-Kermit@CU20B.COLUMBIA.EDU, especially if they have fixes to contribute. The new release is 4E(066). The major changes from 4D(061) (September 86) include: . Support for long packets (but not sliding windows). . Performance improvements: less copying of received data, more efficient i/o, especially when receiving files. . C-Kermit now takes its init file always, even if invoked with command-line action arguments. . Easy escape from packet mode (^C^C at any time). . A file bytesize mask to 'set file type {text, binary} {7, 8}' so that Kermit can be used to strip 8th data bit during file transfer (e.g. of Wordstar files), independent of parity setting. Default 8. . A new 'set terminal bytesize {7, 8}' command. Default 7. . A 'set retry' command to adjust packet retransmission limit. . Support for the Macintosh with Megamax C, so that for the first time Mac Kermit can be built directly on the Mac. . New support Data General AOS, AOS/VS, MV/UX and possibly RDOS & other DG operating systems. . New support for Apollo Aegis. . Continued support for the Commodore Amiga. . New 'make' options for sys5r3, CIE Regulus, HP-UX, IBM IX/370, Zilog, and some others. . Better statistics reporting. . Major bugs fixed: - Loss of trailing control characters at end of file when sending. - 2-character checksum now works with 8-bit binary files. - Background/take-file interaction fixed (maybe?). - Insertion of spurious CRLF at position 4096 when doing 'kermit -k'. - Parsing of multine 'get' command (again). . Many minor bugs fixed. Benchmarks show a slight improvement in efficience when sending files with regular length packets, and large improvement when receiving files, and a very dramatic improvement when receiving files when using parity. The improvements are most noticable on systems where the CPU is the bottleneck. For instance, transferring a 16K text file between a VAX-11/750 and a Rainbow at 9600 baud, using even parity and 94-character packets, the following effective baud rates were observed: C-Kermit Version 4D(061) 4E(066) Send 3500 3920 (a 12% improvement) Receive 800 4223 (a 428% improvement) Use of long packets improves efficiency even more, up to a point (a function of the packet length and the particular system) past which it degrades again. A good length for VAXes seems to be 300-800, where we get effective baud rates in the 50-80% range (provided we have clean lines and no retransmissions) -- higher efficiency at lower baud rates, and even higher in all cases when compression can be done. For instance, the following efficiencies were observed when sending the typical Unix 8K program core image (which has lots of 0's in a row) at 9600 baud from a Rainbow to each of two typically loaded VAXes: ------ VAX 8700 ------- ------- VAX 750 ------- Packet Effective Effective Length Baud Rate Efficiency Baud Rate Efficiency 40 4481 47% 3414 36% 94 6518 68% 4217 44% 200 7170 75% 4780 50% 300 7966 83% 4780 (peak) 50% 500 7966 83% 3773 39% 800 8962 (peak) 93% 2757 29% 1000 7966 83% 2390 25% By the way, a caution to those who are running Ultrix 2.0 on VAX 8700's: Kermit (any version), and probably any program like Kermit, doesn't work very well at 9600 baud on DMZ's with fast PC's like IBM ATs or PS/2s, but does OK at 4800 and below, or at 9600 baud with slower PCs like Rainbows, PC-1's, etc. But Kermit works fine with the same PCs on 750s, 8650s, and other non-BI VAXes. A plea for help with the non-Unix versions: . For all versions, there's been a change to CKxTIO.C (the system-dependent terminal i/o and interrupt procedures for system x) that allow for much more efficient operation with parity; the change is in ttinl(), and I tried to apply it to the various modules, probably incorrectly (and in some, I hadn't the slightest idea what to do). All but the Unix version (ckutio.c) are untested. . The Data General, Apollo, and Amiga support comes from Phil Julian and Jack Rouse at SAS Institute. Their work applied to 4D(061), and I tried to integrate it into the new version. I'm sending them a tape with the new files so they can test it out; meanwhile, if people with Data General systems can try to build from the source and report on the results, that would be great, especially if it still works. . The VMS support hasn't changed, except for the ttinl() business. I have a volunteer who's souping up the VMS support for C-Kermit (in light of the "stable" status of Stevens Kermit-32), and will send him a tape, and hope to get results back in a couple months. . The Macintosh support is a major new change. It now compiles directly on the Mac, under Megamax C, thanks to Jim Noble of Planning Research Corp, who will also get a tape. Again, this support was added to 4D(061), and needs to be rebuilt for 4E(066). If anybody can try this, please report back. And if it works, send in new CKMKER.HQX and CKMKEY.HQX files for 4E. And if anybody wants to try converting it to Apple MPW C, or Lightspeed C, etc, that would be good too. The files are in KER:XK*.* on CU20B.COLUMBIA.EDU (available via anonymous FTP) and XK* * on CUVMA (available via BITNET KERMSRV), and will be on Kermit Tape B, and should also show up at Oklahoma State U for UUCP access within a couple weeks. The new files don't replace the current C-Kermit files (CK*.*), and will not do so until all the systems demonstrably work. In order to use these files, you have to rename them to CK*.* (or ck*.*) so that the various Makefiles and other build procedures work, and the include (.h) files have the right names. There's a program to do this, XKTOCK.C, which should be fairly portable (if it doesn't work, the files can be renamed by hand). Since the collection of files is quite large, you might want to make a judicious selection if obtaining them over networks: Group Size Description KER:CKC*.* 111K Required for all systems. KER:CKW*.* 13K Wart, required for all systems. KER:CKU*.* 545K For Unix, VMS, Data General, Apollo. Includes Unix docs. KER:CKM*.* 488K Macintosh specifics. KER:CKI*.* 96K Amiga specifics. KER:CKD*.* 892K Data General specifics. KER:CKV*.* 67K VAX/VMS specifics. Total size approximately 2.2MB KER:CKP*.* (these files don't exist yet, but "P" is reserved for IBM PC) KER:CKH*.* (not available yet, reserved for Harris, see below) (On BITNET/EARN, leave out the KER: and replace the period by a space.) A detailed list of changes is in the file XKUKER.UPD, and the documentation (CKUKER.MSS, .DOC, .BWR, .NR) has been revised to reflect the new features. Thanks in advance to anyone who is willing to take a shot at evaluating and/or fixing all of this, and apologies for releasing it and then disappearing. And thanks to the many people (listed in the XKUKER.UPD file) who contributed to this release. Finally, if you succeed in building and running this program for a any system at all besides VAXes with Ultrix, please report back the system, OS, OS version, and maybe some particulars like maximum baud rates, best packet size, problems, idiosyncracies, and tricks. - Frank P.S. I just got a tape from David Wilson of the Waisman Center in Madison, WI, with C-Kermit 4D(061) support for Harris computers with DMACPs or CNPs (whatever those are!), but according to his letter, major changes were required to the system-independent (CKC*.*, CKU*.*) modules. It's too late to try to integrate this with the new stuff. Maybe in September. ------------------------------ Date: Tue 15 Sep 87 18:24:28-EDT From: Frank da Cruz Subject: Announcing C-Kermit 4E(067) To: Info-Kermit@CU20B.COLUMBIA.EDU Now that the beta test of version 4E(066) of C-Kermit has had some time to fester, and some good bug reports (and fixes!) have trickled in, it's time to announce a new release, 4E(067). This one only includes fixes for the reported bugs, plus a couple of minor additions. If it checks out OK, it will replace 4D(066) as the standard C-Kermit release. Checking out OK means that it is not inferior to 4D(066) in any way, so that no harm would be done by replacing it. The major changes include: - Fix to allow C-Kermit to run on Pyramid & similar systems. - The wild "set send packet-length" command was tamed. - Allow "set prompt" to work, even from init file. - Problems with packet retransmissions in response to CRLFs should be gone. - Added dial support for the Concord Condor CDS 220 2400b modem. - Changed Xenix compilation options a bit. - New make options for CDC VX/VE, "clean", and "lint". - Set effective group & user IDs on BSD systems for csh command execution. - Fix parsing of "show parameters". - Fix parsing of "remote cwd" from take-command file. - Breakup of source lines longer than 80 characters. - Supply missing functions & symbols for VAX/VMS. - Cosmetic & lint-suggested changes. See the file xkuker.upd for details. The affected files are (just so you don't have to transfer the whole collection again): xkcfns.c, xkcfn2.c, xkcmai.c, xkudia.c, xkufio.c, xkutio.c, xkuusr.h, xkuusr.c, xkuus2.c, xkuus3.c, xkvfio.c, xkvtio.c, xkuker.bwr, xkuker.mak, xkuker.upd There are no new binaries or hex files. People with Unix (any flavor), VAX/VMS, Data General AOS, Macintosh, Apollo, or Amiga systems are urged to get these files and build the new version from source. Anyone who is equipped to build this program from source for the Macintosh or Amiga is further exhorted to do so, then report any bugs (or better still, report if there aren't any!), and if the result is usable, send in the .HQX or .BOO file. If no significant problems are reported with the Unix, VMS, or Macintosh implementations within a few weeks, this will become the standard distributed version of C-Kermit, so that we don't have to carry the CK*.* and XK*.* files side by side, which will make room for some new additions to the "popular minis and mainframes" area (Tape B). Thanks to everyone who sent in bug reports, suggestions, and fixes -- Joe Doupnik, David Herron, Steve Walton, Paul Placeway, S.O. Lidie, Jim Guyton, Markku Toijala, and many others. ------------------------------