Re: patch: under Windows, cvs status reports "memory exhausted"

[Derek Robert Price]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Chris Bohn wrote:

> in cvs 1.12.6 client/server mode
>
> This is also tracked in issue 168:
> http://ccvs.cvshome.org/issues/show_bug.cgi?id=168
>
> I wanted to submit the patch to this list as well because that is the
> way suggested in HACKING.  See the issue for full details, but the
> patch is also pasted below.  I really think this should be in the next
> release since Windows users can't run correctly without it.  The root
> problem is in subr.c, but the problem affects nearly all subcommands
> because the problem code is hit in send_file_names, which is called
> for nearly everything.


I'm not sure what is causing your problem on Windows, but I doubt it is
this.  x2realloc(str, &size) is guaranteed to increase SIZE with each
call.  This particular implementation doubles it and there should be no
differences under Windows.

Can you supply a stack trace for your failure case?

> There is a whole testing framework for unix, but there is nothing for
> Windows.  I think if there was any for Windows, this would have been
> caught because doing a simple cvs status with a file name argument
> exposes the problem.  I realize many of the developers of cvs do their
> work on unix, but if Windows is going to be supported, then some type
> of regression testing framework should probably be made for it too,
> especially now with different code paths because of the case
> sensitivity changes (where this bug was located).


Yep.  This would be nice.  I've thought so for awhile.

Derek

- --
                *8^)

Email: derek@ximbiot.com

Get CVS support at <http://ximbiot.com>!
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.1 (GNU/Linux)
Comment: Using GnuPG with Netscape - http://enigmail.mozdev.org

iD8DBQFAYzzvLD1OTBfyMaQRAhA8AJwNLtEXs8yg5t+8bq+2/64bNCs7dQCgxxpV
eza+QFmOGCPMyWwW4egkGPw=
=dXCe
-----END PGP SIGNATURE-----