[1.11.5]: cvs edit & cvs editors bugs

[Marshall, Simon]

Hi, I sent this email a couple of weeks ago without any response so far,
but I'm not sure this is the correct address to email to.  The original
email doesn't appear in the bug@ccvs.cvshome.org archive under
http://ccvs.cvshome.org/servlets/ProjectMailingListList, but then again
there is nothing after Oct 2002 either.  I can't find any record of
these bugs under http://ccvs.cvshome.org/servlets/ProjectIssues.

On reflection, now I'm pretty sure the first is a bug: CVS doesn't deal
with a user cvs editing a file with the same name in different
directories.  The second seems a no-brainer...

Am I raising these problems in the correct way?

TIA, Simon.

> -----Original Message-----
> From: Marshall, Simon 
> Sent: 02 May 2003 16:43
> To:   'bug-cvs@gnu.org'
> Subject:      [1.11.5]: cvs edit & cvs editors bugs?
> 
> Here are 2 odd problems; I'm not sure what to make of them.  Could I
> have screwed up somehow?  Are they bugs?
> 
> We have a CVS repository module "carma" previously created from an RCS
> repository.  Let's say it contains a trunk plus branches B1 & B2.
> I've done a "cvs watch on carma".
> 
> 
> Problem 1:
> ==========
> 
> User A does "cvs edit Makefile" in A's top-level working directory
> ~A/carma/B1 of a branch B1.
> cvs editors reports:
> 
> Makefile        A        [date]    [host] /users/A/carma/B1
> 
> This is OK.  The fileattr entry says:
> 
> FMakefile
> _watched=;_editors=A>[date]+[host]+/users/A/carma/B1;_watchers=A>tedit
> +tunedit+tcommit
> 
> This seems fine.
> 
> User A does "cvs edit Makefile" in A's top-level working directory
> ~A/carma/B2 of a branch B2.
> cvs editors reports:
> 
> Makefile        A        [date]    [host] /users/A/carma/B2
> 
> What happened to the other edit?  Now the fileattr entry says:
> 
> FMakefile
> _watched=;_editors=A>[date]+[host]+/users/A/carma/B2;_watchers=A>tedit
> +tunedit+tcommit
> 
> Surely this is wrong?  What happened to the entry for branch B1?  Is
> it a CVS bug?  Note that if A does "cvs unedit Makefile" in A's
> top-level working directory ~A/carma/B2, cvs editors report that there
> is no edit for Makefile.  Not surprisingly, as the fileattr says:
> 
> FMakefile _watched=
> 
> So it appears that "cvs edit" erroneously removed entries from
> fileattr when branches are used.  Maybe fileattr needs to be taught
> about branches?
> 
> 
> Problem 2:
> ==========
> 
> User A does "cvs edit Makefile" in A's top-level working directory
> ~A/carma/trunk.
> cvs editors reports:
> 
> Makefile        A        [date]    [host] /users/A/carma/trunk
> 
> This is OK.  The fileattr entry says:
> 
> FMakefile
> _watched=;_editors=A>[date]+[host]+/users/A/carma/trunk;_watchers=A>te
> dit+tunedit+tcommit
> 
> This seems fine.
> 
> User B does "cvs edit Makefile" in B's top-level working directory
> ~B/carma/trunk.
> cvs editors reports:
> 
> Makefile        A        [date]    [host] /users/A/carma/trunk
> 
> Surely this is wrong?  Why doesn't it list user B's edit?  Is it a CVS
> bug?  The fileattr entry says:
> 
> FMakefile
> _watched=;_editors=A>[date]+[host]+/users/A/carma/trunk,B>[date]+[host
> ]+/users/B/carma/trunk;_watchers=A>tedit+tunedit+tcommit,B>tedit+tuned
> it+tcommit
> 
> This seems fine; it contains info for the edit of both users.  Note
> that if user A does "cvs unedit Makefile" in A's top-level working
> directory ~A/carma/trunk, cvs editors reports:
> 
> Makefile        B        [date]    [host] /users/B/carma/trunk
> 
> So it appears that "cvs editors" does not process the fileattr entries
> correctly.
> 
> 
> Please let me know if you need any more information for these
> problems.  Thanks, Simon.
>