On Mon, Mar 14, 2011 at 7:06 AM, Daniel Becroft <djcbecr...@gmail.com>wrote:

> On Sat, Mar 12, 2011 at 1:50 AM, C. Michael Pilato <cmpil...@collab.net>wrote:
>
>> On 03/11/2011 10:03 AM, Arwin Arni wrote:
>> > Index: ../subversion/tests/cmdline/merge_tests.py
>> > ===================================================================
>> > --- ../subversion/tests/cmdline/merge_tests.py        (revision 1080126)
>> > +++ ../subversion/tests/cmdline/merge_tests.py        (working copy)
>> > @@ -16586,6 +16586,102 @@
>> >    if not os.access(beta_path, os.X_OK):
>> >      raise svntest.Failure("beta is not marked as executable after
>> commit")
>> >
>> > +@XFail()
>> > +def dry_run_merge_conflicting_binary(sbox):
>> > +  "dry run merge should not create conflict resolution files"
>>
>> This long description line triggers the AssertionError about the test
>> docstring needing to be 50 characters or less.
>>
>> > +  svntest.actions.run_and_verify_merge(other_wc, '2', '3',
>> > +                                       sbox.repo_url, None,
>> > +                                       expected_output,
>> > +                                       expected_mergeinfo_output,
>> > +                                       expected_elision_output,
>> > +                                       expected_disk,
>> > +                                       expected_status,
>> > +                                       expected_skip,
>> > +                                       None, None, None, None, None,
>> > +                                       True, True,
>> '--allow-mixed-revisions',
>> > +                                       other_wc)
>>
>> As this is a test of a dry-run merge, I find the use of
>> run_and_verify_merge() a bit obfuscating.  I think it'd be better to
>> explicitly run a --dry-run merge so that it's obvious that what you're
>> testing is exactly that.
>>
>> And, as I said elsethread, the patch didn't even apply to HEAD.  So that
>> needs to be reworked.
>>
>
> Hi Mike,
>
> One of the advantages in using run_and_verify_merge() is that if dry_run =
> TRUE, it does it's own check to ensure that the working copy is not
> modified. IMO, this is better than explicitly building the tree prior to the
> merge, and then re-checking the merge.
>
> However, I'm finding that running an explicit merge works, but running
> run_and_verify_merge() does not (conflict files still get created).
>

Never mind, I just found the problem. Using run_and_verify_merge() with
dry_run = True runs both a dry-run and a wet-run update. Since the wet-run
update always gets run, the conflict files always get created.

Cheers,
Daniel B.

Reply via email to