Re: CVS problem with ssh

[Derek Price]

Paul Eggert wrote:

>Yes, I generated this patch:
>
>http://lists.gnu.org/archive/html/bug-cvs/2005-07/msg00034.html
>  
>

Actually, I still had that patch.  In the email I quoted, I thought you
were making reference to a different solution, that didn't rely on a
second process.  I think I have a way to do this, but I thought if you
already had code it might be a head start.  I will fall back on your
forking patch if this idea doesn't work out.

>Thanks.  I'd greatly appreciate it if you could get to the bottom of
>the problem.  (If you can reproduce it, that's probably most of the
>work right there....)
>  
>

The problem is recurring in nightly testing on BSDI, I believe it is,
with the current CVS sources.  It has also cropped up again in some
recent Solaris testing I've been setting up on the SourceForge Compile
Farm.  If we can install a fix and the problem doesn't recur for a week
or two, I'll be fairly satisfied that we got it, but I recall that
removing the select(fileno(std{out,err}), ...) calls in, um, I think
it's handle_m & handle_e in client.c, should make the sshstdio-6 test in
CVS's test suite to fail every time.

>One more thing.  I am still using CVS_RSH=/home/eggert/bin/ssh4cvs in
>my environment, where ssh4cvs is the following script.  Perhaps this
>could be put into the CVS FAQ?  Even if we fix the bug for newer
>  
>

If you'd like to enter it yourself, the new FAQ is Wikied at
<http://ximbiot.com/cvs/wiki/index.php?title=CVS_FAQ>.  Otherwise, I'll
try to make some time to enter it sometime soon.

Regards,

Derek

-- 
Derek R. Price
CVS Solutions Architect
Ximbiot <http://ximbiot.com>
v: +1 717.579.6168
f: +1 717.234.3125
<mailto:derek@ximbiot.com>