[5272] in SIPB bug reports
Re: sipb xdvi
daemon@ATHENA.MIT.EDU (Albert Dvornik)
Sat Apr 1 14:27:22 1995
From: "Albert Dvornik" <bert@MIT.EDU>
To: Eric Mumpower <nocturne@MIT.EDU>
Cc: bug-sipb@MIT.EDU
In-Reply-To: nocturne@mit.edu's message of 30 Mar 1995 (06:05:29 EST).
<9503301105.AA15308@portnoy.MIT.EDU>
Date: Sat, 01 Apr 1995 14:27:07 EST
> What's wrong:
> xdvi crashed, apparently, with this error message:
> xdvi: unknown op-code 255, offset 1916
> [1] + Exit 1 xdvi ~/.3/design.dvi
...
> Please describe any relevant documentation references:
> I was re-running latex on the source LaTeX file from emacs at
> the time. It seems likely that this was related to the crash.
The scenario of what occurred here is as follows.
1. LaTeX starts re-generating the .dvi file.
2. xdvi starts re-reading the xdvi file, because it checks if the
file got changed whenever it gets an expose event.
3. xdvi runs into end of file and realizes it doesn't know what to
do with it. Depending of where in the document this occurs,
xdvi will either exit or complain.
4. LaTeX happily finishes re-generating the .dvi file.
Since there is no real way for xdvi to know if the file is being
modified at the time, the only improvement you could possibly make to
the scheme above is to make xdvi deal more gracefully with the error
condititions. Yes, race conditions suck. =) (Having xdvi just re-read
the file is not a solution, since it would get you stuck if the file
is really corrupt...)
I'm not sure if trying to fix this is really worth the effort; I'd
rather spend time getting a newer version of xdvi into the newtex
locker.
--bert
