Re: CVS update: MODIFIED: src ... (was: join not producing conflict)

[Mark D. Baushke]

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

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

> I found two cases you hadn't checked for and got irked at the way the
> code was behaving.  Let me know what you think.

Nice catch...

...
> |  +
> |  +      # The case where the merge target is up-to-date and its base
> revision
> |  +      # matches the second argument to -j: CVS doesn't bother attempting
> |  +      # the merge since it already knows that the target contains the
> |  +      # change.
> |  +      dotest join6-3.2 "${testcvs} diff temp\.txt" ""

The \. should be . in the above line, unless I am missing something.

Hmmm, given you have just done a commit of temp.txt, the diff should not
be needed as the CVS/Entries timestamp should match the file timestamp
already...

> |  +      dotest join6-3.3 "${testcvs} update -j1.1 -j1.2 temp\.txt" \

The \. should be . in the above line, unless I am missing something.

> |  +"temp\.txt already contains the differences between 1\.1 and 1\.2"
> |  +      dotest join6-3.4 "${testcvs} diff temp\.txt" ""

Ditto.

The rest of the tests look fine.

> |  +    /* FIXME: the noexec case is broken.  RCS_merge could be doing the
> |  +       xcmp on the temporary files without much hassle, I think.  */
> |  +    if (!noexec && !xcmp (backup, finfo->file))

Is it reasonable to add a some "${testcvs} -n ..." test cases?

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

iD8DBQE/mcmd3x41pRYZE/gRAtzqAJ4ow2eE4UPaDx94fFsU8rBQY3X/YwCdEFBq
GFr4/veX+XOW2H755qf7NIc=
=tHYD
-----END PGP SIGNATURE-----