Re: regression in cvs 1.11.1p1, "cvs checkout" command

[Derek R. Price]

Russ Tremain wrote:

> I have found what I believe to be a regression in 1.11.1p1 from 1.11.
>
> The symptom is that the top level directories that are initially
> checked out have a value of the actual directory name in  the CVS/Repository
> meta-file instead of the constant "CVSROOT/Emptydir".  I assume that the
> Emptydir value prevents "cvs update -d" from picking up top-level
> directories that have been explicitly excluded by the initial checkout.
>
> This causes the next "cvs update -d" to bring down all directories
> at the same level that have been explicitly excluded by the initial checkout.
>
> The behavior described above appears when cvs 1.11.1p1 is running in pserver
> mode. cvs 1.11 and 1.10.8 pservers do not have this behavior.  All clients
> (10.8, 1.11, and 1.11.1p1) have the same behavior when 1.11.1p1 is the server.
>
> Please contact me if you need additional details.

This was an intentional change...  from src/ChangeLog:

2000-09-27  Larry Jones  <larry.jones@sdrc.com>

        * commit.c (commit): Don't sleep before returning in server mode,
        just let the client do it.
        * update.c (do_update): Ditto.

        * sanity.sh (find_tool): Correct method of checking for GNU tools.

        * checkout.c (checkout_proc): Match up user directories with
        repository directories instead of using Emptydir.
        * sanity.sh (cvsadm, emptydir): Update to match.

I'm not sure what the compromise is.  Maybe someone has a suggestion?

Some history:  http://www.mail-archive.com/bug-cvs@gnu.org/msg00860.html

Derek

--
Derek Price                      CVS Solutions Architect ( http://CVSHome.org )
mailto:dprice@collab.net         CollabNet ( http://collab.net )
--
My homework was not stolen by a one-armed man.
My homework was not stolen by a one-armed man.
My homework was not stolen by a one-armed man...

          - Bart Simpson on chalkboard, _The Simpsons_