You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@subversion.apache.org by Paul Burba <pt...@gmail.com> on 2013/04/12 16:11:20 UTC
Re: [Issue 2897] Reflective merges are faulty
On Thu, Apr 11, 2013 at 5:15 PM, <ju...@tigris.org> wrote:
> http://subversion.tigris.org/issues/show_bug.cgi?id=2897
>
>
>
> User julianfoad changed the following:
>
> What |Old value |New value
> ================================================================================
> Status|NEW |RESOLVED
> --------------------------------------------------------------------------------
> Resolution| |FIXED
> --------------------------------------------------------------------------------
>
> ------- Additional comments from julianfoad@tigris.org Thu Apr 11 14:15:36 -0700 2013 -------
> Closing as fixed.
>
> Current trunk (which will become Suversion 1.8) supports merging to-and-fro
> between two branches, automatically performing the right sort of merge
> (reintegrate or not, depending on which direction the previous full merge was)
> to take all the not-yet-merged changes from the source branch to the target
> branch. The release-notes description is at
> <http://subversion.apache.org/docs/release-notes/1.8.html#auto-merge>.
>
> It is not perfect. For one thing, it does not deal correctly with revisions
> that have been cherry-pick merged between the branches in all cases (it does in
> some cases). However, it solves the basic requirement.
Hi Julian,
Can we really call this fixed? As you point out it does not deal with
cherry pick (and subtree) merges in all cases. Is there another issue
for the aspects that still don't work?
On a practical note, we have a test associated with this issue that is
set to XFail:
merge_tests.py 49 'avoid repeated merges for cyclic merging'
--
Paul T. Burba
CollabNet, Inc. -- www.collab.net -- Enterprise Cloud Development
Skype: ptburba
Re: [Issue 2897] Reflective merges are faulty
Posted by Paul Burba <pt...@gmail.com>.
On Fri, Apr 12, 2013 at 1:39 PM, Julian Foad <ju...@btopenworld.com> wrote:
> Paul Burba wrote:
>
>> On Thu, Apr 11, 2013 at 5:15 PM, <ju...@tigris.org> wrote:
>>> http://subversion.tigris.org/issues/show_bug.cgi?id=2897
>>>
>>> User julianfoad changed the following:
>>>
>>> What |Old value |New value
>>> =============================================================
>>> Status|NEW |RESOLVED
>>> Resolution| |FIXED
>>>
>>> ------- Additional comments from julianfoad@tigris.org Thu Apr 11 14:15:36
>>> Closing as fixed.
>>>
>>> Current trunk (which will become Suversion 1.8) supports merging to-and-fro
>>> between two branches, automatically performing the right sort of merge
>>> (reintegrate or not, depending on which direction the previous full merge was)
>>> to take all the not-yet-merged changes from the source branch to the target
>>> branch. The release-notes description is at
>>> <http://subversion.apache.org/docs/release-notes/1.8.html#auto-merge>.
>>>
>>> It is not perfect. For one thing, it does not deal correctly with
>>> revisions that have been cherry-pick merged between the branches
>>> in all cases (it does in some cases). However, it solves the basic
>>> requirement.
>>
>> Hi Julian,
>>
>> Can we really call this fixed? As you point out it does not deal with
>> cherry pick (and subtree) merges in all cases. Is there another issue
>> for the aspects that still don't work?
>
> Well, I admit it's a bit bold to call it fixed, but I looked at what the issue was asking for, and it's almost entirely about basic repeated syncs and full merging to-and-fro between a pair of branches. Those merges were the reason for opening the issue, so I think it's fair to close it on those terms.
>
> Of course we want to support cherry picks as well. I think the best thing would be to open a new issue. I don't think there is on, so I'll do that.
Issue #2897 has a lot of discussion not terribly relevant to the
problem, so a new issue sounds find to me.
> But if you feel this "Reflective merges are faulty" issue should stay open, we can re-open it.
>
>
>> On a practical note, we have a test associated with this issue that is
>> set to XFail:
>>
>> merge_tests.py 49 'avoid repeated merges for cyclic merging'
>
> OK, I'll take a look at that. Thanks.
The test is of the criss-cross cherrypick variety. So we can probably
just change the associated issue for this test from #2897 to the new
issue.
--
Paul T. Burba
CollabNet, Inc. -- www.collab.net -- Enterprise Cloud Development
Skype: ptburba
Re: [Issue 2897] Reflective merges are faulty
Posted by Julian Foad <ju...@btopenworld.com>.
Paul Burba wrote:
> On Thu, Apr 11, 2013 at 5:15 PM, <ju...@tigris.org> wrote:
>> http://subversion.tigris.org/issues/show_bug.cgi?id=2897
>>
>> User julianfoad changed the following:
>>
>> What |Old value |New value
>> =============================================================
>> Status|NEW |RESOLVED
>> Resolution| |FIXED
>>
>> ------- Additional comments from julianfoad@tigris.org Thu Apr 11 14:15:36
>> Closing as fixed.
>>
>> Current trunk (which will become Suversion 1.8) supports merging to-and-fro
>> between two branches, automatically performing the right sort of merge
>> (reintegrate or not, depending on which direction the previous full merge was)
>> to take all the not-yet-merged changes from the source branch to the target
>> branch. The release-notes description is at
>> <http://subversion.apache.org/docs/release-notes/1.8.html#auto-merge>.
>>
>> It is not perfect. For one thing, it does not deal correctly with
>> revisions that have been cherry-pick merged between the branches
>> in all cases (it does in some cases). However, it solves the basic
>> requirement.
>
> Hi Julian,
>
> Can we really call this fixed? As you point out it does not deal with
> cherry pick (and subtree) merges in all cases. Is there another issue
> for the aspects that still don't work?
Well, I admit it's a bit bold to call it fixed, but I looked at what the issue was asking for, and it's almost entirely about basic repeated syncs and full merging to-and-fro between a pair of branches. Those merges were the reason for opening the issue, so I think it's fair to close it on those terms.
Of course we want to support cherry picks as well. I think the best thing would be to open a new issue. I don't think there is on, so I'll do that.
But if you feel this "Reflective merges are faulty" issue should stay open, we can re-open it.
> On a practical note, we have a test associated with this issue that is
> set to XFail:
>
> merge_tests.py 49 'avoid repeated merges for cyclic merging'
OK, I'll take a look at that. Thanks.
- Julian