[18923] in Athena Bugs
Re: sun4 8.4.22: rcs under athena solaris
daemon@ATHENA.MIT.EDU (Angie Kelic)
Tue Apr 24 12:38:44 2001
Message-Id: <200104241638.MAA19284@tiki-god.mit.edu>
To: "t. belton" <tbelton@MIT.EDU>
cc: bugs@MIT.EDU
In-reply-to: Your message of "Tue, 24 Apr 2001 12:31:04 EDT."
<Pine.GSO.4.30L.0104241223320.15989-100000@all-night-tool.mit.edu>
Date: Tue, 24 Apr 2001 12:38:41 -0400
From: Angie Kelic <sly@MIT.EDU>
Playing with things some, it seems that it is Solaris diff that
is broken. Running the RCS binaries out of the gnu locker works
(I believe they use gdiff), though the binaries in the gnu locker
are old.
Using my Linux machines (with or without -kb even) allows checkin
of binaries with no problem. (This seems to be the most consistent
and least error prone workaround).
I suspect this is some sort of failure in Solaris diff that would need
to get bug reported to Sun. Perhaps it's fixed in 9.0, or perhaps
a version of RCS that uses gdiff and a version of gdiff could get
placed in /usr/athena/bin for Solaris.
--Angie
>I am surprised this works properly under Linux! I've used RCS on a slew of
>platforms over the years and it was my opinion that it just didn't handle
>binaries well, period, with or without -kb.
>
>I'd be interested to know if someone else has a way of repairing the
>problem Angie encountered (as opposed to preventative measures, like
>"don't use RCS on binaries") that doesn't destroy the version record.
>(Deleting the ,v file by hand and checking the file in freshly solves it,
>but that's less than optimal.)
>
>I ran into this a couple of times last year with some campus-map files;
>the original author had put all the GIFs in RCS and several of them
>flunked checkin in this way when replaced.
>
>On Sun, 22 Apr 2001, Angie Kelic wrote:
>
>> System name: tomcat.mit.edu
>> Type and version: SPARC/5 8.4.22 (with mkserv)
>> Display type: cgsix
>>
>> Shell: /bin/athena/tcsh
>> Window manager: fvwm2
>>
>> What were you trying to do?
>> Check a binary file into RCS. The file was created using
>> rcs -kb and checked out using co -kb so I should have been able
>> to check it back in.
>>
>> What's wrong:
>> When I tried to check the file back in (ci -u) I got the
>> following error
>>
>> >> .
>> Binary files /tmp/deLaa43 and /tmp/T2PGa433 differ
>> ci: RCS/nointel.sav.bin,v: diff failed
>> ci aborted
>>
>>
>> What should have happened:
>> It should Just Work(TM) like it does under linux
>>
>
