Thanks for the nice bug report Senthil. We did not catch this during
development as we didn't have testcase of 'merge -g' for normal
merges(we have one for --record-only case).
Senthil, it would be great if you can convert this to a python test.
You don't need that delete newdir stuff and many files, in fact all we
need is change inside 'subdir' that got added later(on a different
commit from its parent creation commit) than its parent from whom merge
is attempted.
In our greek tree, create A/B1(via svn mkdir) and A/B1/mu(svn add) and
commit this via r2,
Copy A as A_COPY in r3.
Do some change in A/B1/mu at r4.
Merge -g from A to A_COPY.
Committed the fix at r26827.
With regards
Kamesh Jayachandran
Senthil Kumaran S wrote:
> Hi Kamesh,
>
> kameshj@tigris.org wrote:
>> Author: kameshj
>> Date: Thu Sep 27 07:25:29 2007
>> New Revision: 26803
>>
>> Log:
>> Fix issue 2821.
>> Improve merge algorithm used for target trees with child paths which
>> have mergeinfo that differs from their parent.
>
> With the above fix, I think there is a possible bug introduced when we
> merge with some deleted subdir in the working copy. I am attaching a
> shell script which explains this problem along with this mail.
> Previously it used to show "Skipped" for these missing paths, but now
> it errors "File not found".
>
> I feel an XFail test case would be more appropriate to prove this issue.
>
> Thank You.
>
> --
> Senthil Kumaran S
> http://www.stylesen.org/
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Fri Sep 28 15:31:17 2007