[cvsnt] Re: calling cvs client from a hook script

[Daniel Lapolla]

Thanks Bo,

The CVSNT server version I will be using in the target machine is 2.0.38 in
a FreeBSD box (but I can update to 2.0.41 if necessary).

The postcommit hook is ok for my needs, but what about the taginfo? I think
it's like a pre-tag... will it lock the files involved in the tagging
operation?

I'm doing some tests in a WinXP workstation and CVSNT 2.0.41 and the file
being tagged doesn't seems to be locked when my taginfo script runs the cvs
rlog (at least it gets rlogged). Is this a normal behavior thant I can base
my scripts on? Or it's buggy and will be corrected in the next versions?

Thanks again,

Daniel

"Bo Berglund" <bo.berglund at telia.com> escreveu na mensagem
news:h6q2c0t3v26gdbr3m0bk0u3df3qfo1mpa1 at 4ax.com...
> On Fri, 4 Jun 2004 18:17:42 -0300, "Daniel Lapolla"
> <ldlapolla at yahoo.com> wrote:
>
> >Hi,
> >
> >I'm thinking about creating a taginfo and a commitinfo hook scripts that,
> >among other things, runs a cvs rlog on the tagged/commited files.
> >
> >Is this an advisable practice? I'm concerned if this practice can end up
> >creating a deadlock ou some other problem.
> >
> >Thanks in advance,
> >
> >Daniel
> >
>
> I think the affected files have been locked by CVSNT at this point in
> time and your cvs operations from within the script on the same module
> will probably fail because of this....
> To my knowledge the only script that executes after the locks have
> been released again is the postcommit script.
> And you have to be careful of the CVSNT version if you use that
> because in some versions the locks were not released. I use postcommit
> to update a website on commits and got the lock error until I updated
> the CVSNT server version.
>
>
> /Bo
> (Bo Berglund, developer in Sweden)