Re: Tree conflict bugs I've just been finding

Neels J. Hofmeyr wrote:
>
> Stefan Sperling wrote:
>> On Thu, Nov 13, 2008 at 07:49:20PM +0000, Julian Foad wrote:
>>> Bug: "resolved" complains if the target's parent isn't a WC:
>>> [[[
>>> $ svn resolved wc2
>>> svn: warning: '.' is not a working copy
>>>
>>> $ svn st wc2
>>> M C wc2/foo
>>> ]]]
>> Hey Julian,
>>
>> I've tried to take a shot at this one. My current patch is below.
>>
>> 4 tests in tree_conflicts_tests.py are failing with this patch:
>> FAIL: tree_conflict_tests.py 9: merge file: modify onto not-file
>> FAIL: tree_conflict_tests.py 11: merge file: del/rpl/mv onto not-file
>> FAIL: tree_conflict_tests.py 13: merge dir: modify onto not-dir
>> FAIL: tree_conflict_tests.py 15: merge dir: del/rpl/mv onto not-dir
>>
>> I don't exactly know why. I may have an error in my patch.
>> Maybe you can investigate?
>>
>> Thanks,
>> Stefan
>
> Although I'm not Julian, I know what the problem is: If we have a
> tree-conflict victim that is unversioned as the target, the patch thinks
> that that victim is a working copy root and skips checking it altogether.
>
> Maybe we ought to rather check whether the parent of the target is a working
> copy, nevermind checking whether the target is a working copy root.

I solved it differently in the end, but it's fixed in r34186. Thanks Stefan!
Wow, I took quite a detour trying to fix it. Anyway, now it's fixed.

I also fixed a case where a tree-conflict victim wasn't visited by
"resolved": a dir that had been deleted, but became a tree-conflicts victim
later on (during merge). I'm guessing that it doesn't give false positives.

~Neels