Re: problem

[Derek R. Price]

Benoît Rouits wrote:

> traces of it from the repository. If at some later time the
> file is re-added, the
> Attic/filename,v with dead state is moved to filename,v and
> given an exists state in the CVS repository. Now because MS
> Windows is somewhat
> case insensitive, a problem can arise if the re-added file is

[. . .]


> I have this problem on M$-Window$ and would like to modify cvs
> sources to fix that.
> I tried to modify add.c but it doesn't work yet. I have 2
> Questions :
> 1. is it technically possible ?

It's debatable whether only the add command should be case sensitive or
whether the rcs library call which actually performs the ressurection
should be made case-insensitice too, but yes, I think so.


> 2. what source code belongs to cvs server, what source code
> belongs to client ?

if (server_active) {} bracketed code is server-only.  if
(!server_active) {} bracketed code is client only.  otherwise, unless it
is code only called from a bracketed area it is shared.

It might help to understand the architecture here:

    cvs --allow-root=... pserver

effectively accepts a socket connection, creates a mirror of the users
workspace in /tmp, then executes the same command the user did in the
/tmp workspace (really a child process), then sends the output back to
the client.  Hmmm...  well, if that's not exact, it's close.

client.c and server.c are good places to look for some client and server
specifc code.

You probably saw, since he was replying to you, but Brad Chisholm gets
technical about a fix in a very recent email, at the least.

    http://www.mail-archive.com/bug-cvs%40gnu.org/msg01177.html



Hope all this helps.

Derek

--
Derek Price                      CVS Solutions Architect ( http://CVSHome.org )
mailto:dprice@collab.net         CollabNet ( http://collab.net )
--
#! perl
@a = ( 0x2E805,0x6B39,0x15B3,0x45993,0x153C,0x1D9F );
for ( @a ) { ( $s, $i )=( 'a', 0 ); $s++ while $i++ < $_; print "$s" }