Date: Sat, 07 Feb 1998 11:21:24 -0800 From: John Polstra <jdp@polstra.com> To: scottm@cs.ucla.edu Cc: current@FreeBSD.ORG Subject: Re: cvs bug (and fix) Message-ID: <199802071921.LAA18068@austin.polstra.com> In-Reply-To: <199802062246.OAA03856@mordred.cs.ucla.edu> References: <199802062246.OAA03856@mordred.cs.ucla.edu>
next in thread | previous in thread | raw e-mail | index | archive | help
In article <199802062246.OAA03856@mordred.cs.ucla.edu>,
Scott Michel <scottm@cs.ucla.edu> wrote:
> This one's been driving me batshit for the last couple of days. I'm
> surprised no one else has encountered it yet.
>
> Line 6016 in /usr/src/contrib/cvs/src/rcs.c needs to be changed from
>
> if (n == EOF)
>
> to
>
> if (n == EOF || num == NULL)
>
> Otherwise you get spurious assert's in findnode() a little later on.
It looks like only an invalid RCS file could cause that to happen.
(Which of course doesn't make it any less of a bug.) Is that right?
> Can someone change, approve, and check this in? TIA
The fix doesn't look quite right to me. A revision number was
expected, but something else (not EOF) was found instead. Shouldn't
it report an error?
How about something like this:
if (n == EOF)
{
/* If n == EOF and num == NULL, it means we reached EOF
naturally. That's fine. */
if (num == NULL)
return NULL;
else
error (1, 0, "%s: unexpected EOF", rcs->path);
}
else if (num == NULL)
error (1, 0, "%s: file contains invalid revision number", rcs->path);
--
John Polstra jdp@polstra.com
John D. Polstra & Co., Inc. Seattle, Washington USA
"Self-knowledge is always bad news." -- John Barth
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199802071921.LAA18068>