Re: cvs add (resurrect) incompatibilities

[Mark D. Baushke]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Derek Robert Price <derek@ximbiot.com> writes:

> Mark D. Baushke wrote:
> 
> >  2004-10-21  Mark D. Baushke  <mdb@cvshome.org>
> >  
> >      * add.c (add): Pay attention to cvswrite mode when resurrecting a
> >      file.
> >      (Report from Frank Hemer <frank@hemer.org>.)
> >      * sanity.sh (resurrection): Add new tests to deal with read-only
> >      mode on a cvs add durring a resurrection.
> 
> 
> I disagree with this patch.  The file shouldn't be marked read-only
> until after commit - it's a modified file and should be in a state the
> user can still edit until commit.

Agreed. That is what the patch does.

> Of course, it should be checked out read-only if other criteria would
> have caused that on up/co when the resurrection is of an uncommitted
> removal (since the resurrected file immediately assumes "unmodified"
> and "unedited" status),

Again, agreed. That is what the patch does.

> but not when the resurrection still needs to be committed as in the
> case you appear to be testing.

Yup... perhaps I need a better ChangeLog entry?

2004-10-21  Mark D. Baushke  <mdb@cvshome.org>

    * add.c (add): Pay attention to cvswrite mode when resurrecting a
    file that was not yet committed.
    (Report from Frank Hemer <frank@hemer.org>.)
    * sanity.sh (resurrection): Add new tests to deal with read-only
    mode on a cvs add durring a resurrection. Verify that -r is not
    honored when the resurrected file must be committed to be seen.


Detailed description:

The change to add.c deals with the case of 

    cvs rm -f file1
    cvs -r add file1

that is, the file already existed in the repository and no change is
required, we may therefore honor the -r request.

The change to sanity.sh tests two things...

    a) Should a ressurection honor the read-only flag? Answer: no.

       cvs rm -f file1
       cvs ci file1
       cvs -r add file1
       test -w file1 || echo there is a bug here

       Implementation: Modify resurrection-6 to add the '-r' switch
       Add resurrection-6b to verify that the file has the write mode
       bit set.

    b) Should a ressurection of a file that has a top-of-branch version
       honor the read-only flag? Answer: no

       cvs rm -f file1
       cvs -r add file1
       test -w file1 && echo there is a bug here

       Implementation: add tests resurrection-{13,14,15} to verify that
       the newly re-added file honors the '-r' switch.

I think that covers it. If there are no other objections, I'd like to
commit this patch.

        -- Mark
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.3 (FreeBSD)

iD8DBQFBeAiRDsmuPPFOO2cRAsdFAJ9HfW2dSOXqRsAFQcvGN2uX/XP3hQCeLCrS
zoGGlzJ2eco/XlwMyHErAog=
=nMSA
-----END PGP SIGNATURE-----