Re: Accessing CVS files during commit (via a script)

[Mark McDonald]

Output from a trigger script:

#!/usr/bin/perl

use Cwd;
use strict;

use Carp;

my $pwd = cwd();

open (F, ">/tmp/junk/test.txt")  or croak ('can not
open /tmp/junk/test.txt');

print F "PWD: $pwd\n";
print F map { "$_\n"} @ARGV;
print F `ls -Rla`;
##########################################

## Output of this script:


PWD: /home/mark/Test/FormProcessor
/cvsroot/FormProcessor
form.html
.:
total 48
drwxr-xr-x    4 mark     mark         4096 Jul 16
12:36 .
drwxr-xr-x    3 mark     mark         4096 Jul 13
14:28 ..
drwxr-xr-x    2 mark     mark         4096 Jul 16
12:36 CVS
drwxr-xr-x    3 mark     mark         4096 Jul 13
14:28 Data
-rwx------    1 mark     mark         2866 Jul 16
12:36 form.html
-rwx------    1 mark     mark         2861 Jul 13
15:41 form.html~
-rwx------    1 mark     mark         9403 Mar 17
07:45 form_processor.cgi
-rwx------    1 mark     mark         8196 Mar 17
07:45 form_test.html

./CVS:
total 20
drwxr-xr-x    2 mark     mark         4096 Jul 16
12:36 .
drwxr-xr-x    4 mark     mark         4096 Jul 16
12:36 ..
-rw-r--r--    1 mark     mark          152 Jul 13
15:41 Entries
-rw-r--r--    1 mark     mark           14 Jul 13
14:28 Repository
-rw-r--r--    1 mark     mark            9 Jul 13
14:28 Root

./Data:
total 32
drwxr-xr-x    3 mark     mark         4096 Jul 13
14:28 .
drwxr-xr-x    4 mark     mark         4096 Jul 16
12:36 ..
-rwx------    1 mark     mark           58 Feb 27
08:42 .cgi
-rwx------    1 mark     mark        12288 Feb 27
08:42 .form.html.cgi.swp
drwxr-xr-x    2 mark     mark         4096 Jul 13
14:28 CVS
-rwx------    1 mark     mark          377 Feb 27
08:42 form.html

./Data/CVS:
total 20
drwxr-xr-x    2 mark     mark         4096 Jul 13
14:28 .
drwxr-xr-x    3 mark     mark         4096 Jul 13
14:28 ..
-rw-r--r--    1 mark     mark          144 Jul 13
14:28 Entries
-rw-r--r--    1 mark     mark           19 Jul 13
14:28 Repository
-rw-r--r--    1 mark     mark            9 Jul 13
14:28 Root

--- Mark <address@hidden> wrote:
> 
> I just did some testing, and modifying the server
> tmp copies is a real bad
> idea, like Dennis said. But it is good that CVS
> makes the to-be-commited files
> available (whether or not by design) to the trigger
> scripts.
> 
> CVS doesn't have precommit triggers and I
> (personally) don't like tools that
> have client side triggers.
> 
> If you are trying to enforce coding syntax/policy,
> have your trigger do want it
> needs to do on a copy of the server tmp files, then
> compare the resultant files
> the tmp files server will commit. If they are any
> differences, reject the
> commit and tell the user to run the beautifier prior
> to commit.
> 
> Mark






Thanks Mark. Ok, then  _where_ are these tmp files?!!
I have basically looked all over my system including
all cvsroot dirs and the current working dir that the
trigger script runs.

I wrote a quick program to do what you said and
printed above...

Thanks in advance.

-Mark McDonald

> 
> --- Dennis Jones <address@hidden> wrote:
> > Mark,
> > 
> > Is this really a good idea?  If you do what you
> are proposing, after you
> > commit a file, changes will be made to the file
> (unbeknownst to you) and I
> > suspect your local copy will immediately be out of
> date, even though you
> > just committed it!  So, a question I have is, will
> CVS *know* that the file
> > is out of date?  I don't know how CVS works in
> this regard...does it get a
> > copy of the file from the repository immediately
> following a commit?
> > Perhaps a CVS expert can shed some light on this
> question.
> > 
> > CVS must do *something* at this point though,
> because it has to modify any
> > keywords you have in your file (such as $Id) after
> the commit, but I suspect
> > that they are modified directly by the client.  If
> this is the case, then
> > you will almost surely have trouble with files
> being out of date and CVS not
> > knowing that this is the case.  It is also
> possible that an update might not
> > even help -- if CVS thinks the file is already
> updated!
> > 
> > So, my next question is, why not run the code
> through the beautifier PRIOR
> > to the commit?  Then you will be sure your files
> are up to date after the
> > commit, and anyone else that checks out (or
> updates) the file will get the
> > beautified copy.
> > 
> > - Dennis
> > 
> > ----- Original Message -----
> > From: "Mark McDonald" <address@hidden>
> > To: "Mark" <address@hidden>; <address@hidden>
> > Sent: Monday, July 16, 2001 8:39 AM
> > Subject: Re: Accessing CVS files during commit
> (via a script)
> > 
> > 
> > > Hi Mark,
> > >
> > > Thanks for the response. I know how to get the
> > > variables though, what I am trying to do is
> actually
> > > open up the files that are being committed.
> > >
> > >  So for example, suppose you wanted to run a
> code
> > > beautifier on the script/file being committed
> within
> > > cvs. You would then need access to the incoming
> file
> > > that cvs has just received.
> > >
> > > I guess what I am asking is if this is possible.
> The
> > > 'Open Source Development Book with CVS' claims
> that
> > > you can do this --but gives no examples as to
> how to
> > > do it.
> > >
> > > Any help is totally appreciated!
> > >
> > > -Mark McDonald
> > >
> > >
> > > # perl this_script  /cvsroot/mod_perl_config
> > > startup.perl
> > > #                            |                  
> |
> > > #                      the CVSROOT/module     
> the
> > > modified file
> > > #
> > >
> > > # perl this_script
> > > /cvsroot/mod_perl_config/db_connections
> > > mikey.connection web_dev.connection
> > > #                             |
> > >           |                  |
> > > #                         the module/sub-dir
> > >        mod file 1         mod file 2
> > >
> > >
> > > --- Mark <address@hidden> wrote:
> > > >
> > > > Ask CVS....
> > > >
> > > > create a trigger script to ask questions and
> find
> > > > out the trigger's runtime
> > > > environemnt...
> > > >
> > > > things to include might be...
> > > >
> > > > print "@ARGV";
> > > > print "$0,$1,$2.....";
> > > > pwd;
> > > > ls -Rla;
> > > > env;
> > > >
> > > > that's how I found out (with pserver anyhow)
> > > >
> > > > Mark
> > > >
> > > > --- Mark McDonald <address@hidden>
> wrote:
> > > > > Does anyone know how this works?!
> > > > >
> > > > >
> > > > > > I am using the CVS Open Source Development
> book
> > > > as a
> > > > > reference to make a
> > > > > > script that will run upon a cvs commit.
> > > > Basically
> > > > > what
> > > > > > I am trying to do is open up the file that
> is
> > > > being
> > > > > > committed to a repository.
> > > > > >
> > > > > > On page 132 of the CVS book (in the top
> note
> > > > > section)
> > > > > > this book states that '
> > > > >
> > > > >   the file names passed to the program
> > > > > > do not refer to the RCS files --they point
> to
> > > > normal
> > > > > > files, whose contents are exactly the same
> as
> > > > the
> > > > > > working-copy files being committed. The
> only
> > > > unusual
> > > > > > aspect is that CVS has them temporarily
> places
> > > > > inside
> > > > > > the repository, so they'll be available to
> the
> > > > > > programs running on the machine where the
> > > > repository
> > > > > > is located.'
> > > > > >
> > > > > >
> > > > > > I have basically tried everything to
> locate
> > > > these
> > > > > > files so I can do things with them. In
> this case
> > > > I
> > > > > am
> > > > > > trying to copy the contents to webservers
> on
> > > > this
> > > > > and
> > > > > > other networks.
> > > > > >
> > > > > > I have check the news groups and CVS's
> homepage
> > > > > > in search of this info. Perhaps you could
> give
> > > > me a
> > > > > > tip!
> > > > > >
> 
=== message truncated ===


__________________________________________________
Do You Yahoo!?
Get personalized email addresses from Yahoo! Mail
http://personal.mail.yahoo.com/