Réf. : Re: [cvsnt] verifymsg (syntax and tests

[feymard at stago.fr]

> On 02/01/2006 14:20 bo.berglund wrote:
>
> On Mon, 2 Jan 2006 11:19:30 +0100, feymard at stago.fr wrote:
>
> >Anyone has a solution?
> >
>  1. (the best solution)
> Educate your users to *always* enter log messges describing what they
> are checking in and why.

It's a quality procedure requirement :(

> 2. (if 1 does not work)
> Use the CVSROOT/verifymsg script to check the log message and write a
> feedback message to the user if it does not comply to the standard.
> Then exit with a non-zero code that makes the commit fail.

Exactly what I want

> As explained in the CVSNT documentation you can have the following in
> the verifymsg file:
>
> DEFAULT <path to log message checking program> %l

Damned, I miss the %l... oupsss...

> Your program (you will have to write it yourself) will then be
> started by CVSNT when there is a commit and the full path to a file
containing
> the contents of the log message is supplied to your checking program
> as a command line argument.

I put a small script on the server, and it's read. But not executed!

with just one line, here is the log windows of Tortoise:

C:\CVSTemp\cvs-serv1368>exit /B 1
Checking in foo.txt;
/repotest01/module01/foo.txt,v  <--  foo.txt
new revision: 1.4; previous revision: 1.3
done

(all lines of my script will be written like this "exit /B 1)
the directory C:\CVSTemp\cvs-serv1368 does not exist.

Other question: where does the log is stroed when verifymsg is called? Have
I to work with stdin?

> But if all you do is to check that there is "something" entered then
> I'm afraid that you will be disappointed because you will probably
> find a lot of dfdfdg or asdr or qwerty or similar meaningless words
> there...

I know that, some project managers make to me this request, so...
They will have to verify the comments. In a next step, when I understand
whole
of veryfymsg, rcsinfo, and others, I could use rcsinfo to force the content
of
comments ;)