On Thu, Jan 22, 2009 at 8:51 AM, Stefan Sperling <stsp_at_elego.de> wrote:
> Hey Paul,
> I know you're working on this, too, but I just to let you know that
> I quickly tested my theory for the fix and it seems to work.
> See the diff below.
> This fixes the non-interactive case (--accept postpone), and the
> interactive case still works. Tested with my reproduction script.
On Thu, Jan 22, 2009 at 8:57 AM, Stefan Sperling <stsp_at_elego.de> wrote:
> Note also that I didn't have time to run make check yet, so I don't
> know yet if this breaks anything else...
My patch and log message is attached, here is the log:
Fix issue #3354 when there is no interactive conflict resolution.
(maybe_resolve_conflicts): Consider no interactive conflict resolution
as analogous to postponing conflict resolution.
(test_list): Remove XFail from update_moves_and_modifies_an_edited_file.
I ran the test suite [FSFS] x [RA_LOCAL] using both of our patches and
all the tests pass, including update_tests.py 52 with the changes I
just made in r35388 (see below).
We are doing essentially the same thing, but I think mine is slightly
clearer (but that's probably because I wrote it :-) Of course,
speaking of clear within the context of an undocumented function is a
bit comical. I'll defer to you all who have spent a lot more time in
this code as to which approach is better.
> update test 52 still has a problem with this patch: It complains that
> alpha.moved.copied was missing from the actual tree on disk.
> Either the expected disk status is incorrect, or the patch below is not
> all that's needed.
It's the test that is incorrect. It forgot to account for the three
unversioned conflict files:
trunk>svn st update_tests-52.other
A + C update_tests-52.other\A\B\E\alpha
> local edit, incoming delete upon update
I was not familiar with the conflict file
update_tests-52.other\A\B\E\alpha.moved.copied. This was added with
the fix for issue #2986 in r27377.
Fixed the test expectations in r35388.
Received on 2009-01-22 18:34:51 CET