Re: [zackw@stanford.edu: cvs/ssh random observation]
Martin,
On Thu, 25 Jan 2001 18:59:21 +0900 (JST), Martin Buchholz <martin(a)xemacs.org>
wrote:
>>>>> "OP" == Ovidiu Predescu
<ovidiu(a)xemacs.org> writes:
OP> Since we are on this topic, I noticed that using CVS over SSH is very
OP> unreliable at cvs.xemacs.org. The GCC people are using this method for several
OP> years now and it's been very reliable for them and certainly for me. I'm
OP> talking about files which are not checked-out/updated, sudden disconnects from
OP> the remote CVS server etc. Martin mentions this on the Web site. What is wrong
OP> with the cvs.xemacs.org setup that makes it unreliable?
Set up your CVS so that all CVS operations that are r/o are done using
the anonymous :pserver:. Then, when you want to do some r/w thing,
use "cvs-rw" instead of "cvs". This has a good chance of making
your
cvs xemacs operations more reliable.
I've been using the method you describe here, I saw it on the XEmacs' CVS Web
page before I asked the question.
My point was that the GCC people are using :ext: with ssh for many, many years,
with no reliability problems, both for checking out and committing. They don't
need a cvs-rw script because their CVS works very well with :ext: and ssh.
My question was what could be different in the XEmacs CVS from the GCC CVS that
makes it so unreliable?
The script below will need modification for use by non-martins...
[script removed]
Thank you very much for this script, I'll modify and use it for the time being.
Best regards,
--
Ovidiu Predescu <ovidiu(a)cup.hp.com>
http://orion.nsr.hp.com/ (inside HP's firewall only)
http://www.geocities.com/SiliconValley/Monitor/7464/