Date: Sun 24 Mar 85 23:14:41-PST From: Bob Larson Subject: Prime kermit bug To: info-kermit@CU20B.ARPA Prime kermit is quoting the 8-bit character even when 8-bit quoting is not being done. Using defaults, this causes ampersands ('&') to be received as lower case f's by some versions of kermit. (I.e. the old unix kermit that is the base of the current os9 kermit effort. Bob Larson ------------------------------ Date: 14-OCT-1985 10:32:48 From: SYSKERMIT%vax1.central.lancaster.ac.uk@ucl-cs.arpa To: INFO-KERMIT Subject : PRIME KERMIT bug query Seems to be a bug in PRIME Kermit so that in server mode it responds to an Init-info (I) packet with a NAK instead of (presumably) an Error packet. Has anyone got a fix for this? Alan ------------------------------ Date : 10 January 1986 From : Rick Burne, Ealing College of Higher Education, London UK Subject : Bug in Prime KERMIT in server mode In server mode PRIME KERMIT responds to an I packet with a NAK rather than anything sensible such as a qualified ACK or an error packet. This should be fixed in the next release from The Source, but in the meantime here is a quick edit to make it respond with an error packet. File MSG_TYPES.PLP Around line 24, insert a line MSG_INIT by 'I' /* Init-info packet */ after the line MSG_EOF by 'Z' /* End of file (EOF) */ File SERVER.PLP Around line 225, insert a new clause in the main do while loop: when (MSG-INIT) do; snd_msg = 'Unimplemented server command'; call send_packet('E',length(snd_msg),msg_number); end; before the line end; /* select */ File REC_MESSAGE.PLP Around line 72, insert a line msg_init, before the line msg_rcv_init) return ('1'b); ------------------------------ 24-Oct-86 04:06:03-EDT,3512;000000000001 Return-Path: <@MIT-MULTICS.ARPA,@BRL-AOS.ARPA:SYSKERMIT%vax1.central.lancaster.ac.uk@cs.ucl.ac.uk> Received: from MIT-MULTICS.ARPA by CU20B.COLUMBIA.EDU with TCP; Fri 24 Oct 86 04:05:45-EDT Received: from BRL-AOS.ARPA by MIT-MULTICS.ARPA TCP; 24-Oct-1986 03:37:45-edt Received: from ucl-cs.arpa by AOS.BRL.ARPA id aa22123; 23 Oct 86 20:02 EDT Received: from vax1.cs.ucl.ac.uk by mv1.Cs.Ucl.AC.UK via Ethernet with SMTP id ad03590; 22 Oct 86 9:56 WET Received: from 44d.cs.ucl.ac.uk by vax1.Cs.Ucl.AC.UK with SMTP id a001317; 22 Oct 86 9:53 BST Received: from vax1.central.lancaster.ac.uk by 44d.Cs.Ucl.AC.UK via Janet with NIFTP id a003664; 22 Oct 86 9:51 BST Date: 22-OCT-1986 09:12:56 From: SYSKERMIT%vax1.central.lancaster.ac.uk@cs.ucl.ac.uk To: INFO-KERMIT <@cs.ucl.ac.uk,@MIT-MULTICS.ARPA:INFO-KERMIT@cu20b.ARPA> Subject: Bug in Prime Kermit 7.57 From: 21-OCT-1986 18:39 To: SYSKERMIT Subj: Date: Tue, 21 Oct 86 18:35:15 GMT To: syskermit@UK.AC.LANCASTER.CENTRAL.VAX1 From: BROOKS@UK.AC.EXETER.PC Subject: Prime Kermit 7.57 Message-ID: There is a bug in PR1ME Kermit version 7.57. If, when in SERVER mode, the user logs off with a BYE command on the micro-Kermit, the NEXT user to get that PRIMOS usernumber can have problems related to the PRIMOS KILL and ERASE characters. It showed up in the Sheffield Editor for us. The problem arises if other software expects that the PRIMOS routine ERKL$$ returns leading zeros in the words that give the user's KILL and ERASE characters as documented in DOC3621-190 Subroutines Reference Guide page 10-23. What is not documented is that when writing the KILL and ERASE the leading byte of the words should be ZERO. This is the only reason that reading returns leading zeros! Unfortunately, PRIMOS Kermit uses leading spaces when writing these values. This in itself causes no problem as when Kermit exits SERVER mode in all ways EXCEPT when it receives a BYE command, it restores the user's KILL and ERASE characters with words it had read previously. Again this seems no problem as LOGO$$ restores the System default KILL and ERASE characters so everything appears OK to the next user. But logo$$ does not seem to alter the leading byte of the words holding KILL and ERASE, hence the problem. The simplest fix in PRIMOS Kermit is as follows:- In GENERIC_CMD.PLP, before call logo$$(0,0,' ',0,0,code); insert the line call xfer_mode(0,code); /* restore terminal characteristics */ Ideally, the first call to erkl$$ in XFER_MODE.PLP should be fixed so that it has leading zeros in the character strings it passes, but this needs more changes to be made in the source. The fix given is short and it works! There may be a problem with forced logouts which this fix obviously won't cope with. The QUIT handler calls xfer_mode so there is no trouble there. This caused us no end of trouble to track down to Kermit! It seemed an epidemic had struck as more and more users hit this problem. (The PRIMOS command TERM -kill and TERM -erase fixes a user in trouble). It didn't help that we changed from Rev 19.3 to Rev 19.4 at the same time as we released Kermit 7.57. Neil Brooks University of Exeter Computer Unit