You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@subversion.apache.org by Julian Foad <ju...@btopenworld.com> on 2008/10/18 12:39:52 UTC
Regression in merge - yank release 1.5.3? [was: svn commit: r33696
- branches/1.5.x]
(Changing the subject line.)
On Sat, 2008-10-18 at 05:54 -0500, Hyrum K. Wright wrote:
> Greg Stein wrote:
> > Should we yank 1.5.3 and publish a 1.5.4 right away?
> >
> > If this is a regression, then it seems we ought to pop out a 1.5.4 right away.
>
> Fine by me, if others feel that we need to. (Don't stay silent on this, people,
> I'd like a few +1's before going through the effort here.)
>
> -Hyrum
From the 1.5.x STATUS file:
> * r33693
> Fix merge bug introduced in the backport of the r29969, r32463, r33013,
> r33016, r33022, r33112 group in 1.5.3.
> Notes:
> There is a minor whitespace conflict in merge.c.
> Justification:
> This bug is a regression in 1.5.3, breaks mergeinfo aware merges when
> ecountered, and has already been seen a couple times in our own repos,
> see http://subversion.tigris.org/servlets/ReadMsg?list=dev&&msgNo=144187
> and http://subversion.tigris.org/issues/show_bug.cgi?id=3294#desc1.
> Votes:
> +1: pburba
>
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: Regression in merge - yank release 1.5.3? [was: svn commit: r33696
- branches/1.5.x]
Posted by "Hyrum K. Wright" <hy...@mail.utexas.edu>.
Paul Burba wrote:
> On Sat, Oct 18, 2008 at 8:57 PM, Arfrever Frehtes Taifersar Arahesis
> <ar...@gmail.com> wrote:
>> 2008-10-18 16:13:34 Mark Phippard napisał(a):
>>> I did not get the impression the bug was THAT bad. It sounds like you
>>> are going by the wording of the backport proposal which makes it sound
>>> like merge is completely broken. I am not against a 1.5.4 but I'd
>>> like to hear from Paul. Perhaps we ought to just move up the 1.5.4
>>> schedule so that it happens in early November?
>> +1 for tagging 1.5.4 circa 2008-11-10.
>>
>>> On Sat, Oct 18, 2008 at 8:39 AM, Julian Foad <ju...@btopenworld.com> wrote:
>>>> (Changing the subject line.)
>>>>
>>>> On Sat, 2008-10-18 at 05:54 -0500, Hyrum K. Wright wrote:
>>>>> Greg Stein wrote:
>>>>>> Should we yank 1.5.3 and publish a 1.5.4 right away?
>>>>>>
>>>>>> If this is a regression, then it seems we ought to pop out a 1.5.4 right away.
>>>>> Fine by me, if others feel that we need to. (Don't stay silent on this, people,
>>>>> I'd like a few +1's before going through the effort here.)
>>>>>
>>>>> -Hyrum
>
> Thinking about this a bit more, I'd like to see 1.5.4 as soon as possible. Why?
Sounds good. We'll roll on Wednesday.
All, please review and vote on STATUS items before then.
-Hyrum
Re: Regression in merge - yank release 1.5.3? [was: svn commit: r33696 - branches/1.5.x]
Posted by Paul Burba <pt...@gmail.com>.
On Sat, Oct 18, 2008 at 8:57 PM, Arfrever Frehtes Taifersar Arahesis
<ar...@gmail.com> wrote:
> 2008-10-18 16:13:34 Mark Phippard napisał(a):
>> I did not get the impression the bug was THAT bad. It sounds like you
>> are going by the wording of the backport proposal which makes it sound
>> like merge is completely broken. I am not against a 1.5.4 but I'd
>> like to hear from Paul. Perhaps we ought to just move up the 1.5.4
>> schedule so that it happens in early November?
>
> +1 for tagging 1.5.4 circa 2008-11-10.
>
>> On Sat, Oct 18, 2008 at 8:39 AM, Julian Foad <ju...@btopenworld.com> wrote:
>> > (Changing the subject line.)
>> >
>> > On Sat, 2008-10-18 at 05:54 -0500, Hyrum K. Wright wrote:
>> >> Greg Stein wrote:
>> >> > Should we yank 1.5.3 and publish a 1.5.4 right away?
>> >> >
>> >> > If this is a regression, then it seems we ought to pop out a 1.5.4 right away.
>> >>
>> >> Fine by me, if others feel that we need to. (Don't stay silent on this, people,
>> >> I'd like a few +1's before going through the effort here.)
>> >>
>> >> -Hyrum
Thinking about this a bit more, I'd like to see 1.5.4 as soon as possible. Why?
A) This *is* a regression from 1.5.2.
B) While merge isn't completely broken, it's fairly simple to trigger
this problem: Rename a directory in trunk and try to merge that change
to any of our branches, you'll get a "svn:
'/repos/svn/!svn/bc/some/branch/path' path not found" error.
C) There is a workaround for B, perform the merge with
--ignore-ancestry and the required revision range(s) and then repeat
the merge with --record-only and the same ranges. But determining the
"required revision range(s)" can be quite cumbersome.
Paul
Re: Regression in merge - yank release 1.5.3? [was: svn commit: r33696 - branches/1.5.x]
Posted by Arfrever Frehtes Taifersar Arahesis <ar...@gmail.com>.
2008-10-18 16:13:34 Mark Phippard napisał(a):
> I did not get the impression the bug was THAT bad. It sounds like you
> are going by the wording of the backport proposal which makes it sound
> like merge is completely broken. I am not against a 1.5.4 but I'd
> like to hear from Paul. Perhaps we ought to just move up the 1.5.4
> schedule so that it happens in early November?
+1 for tagging 1.5.4 circa 2008-11-10.
> On Sat, Oct 18, 2008 at 8:39 AM, Julian Foad <ju...@btopenworld.com> wrote:
> > (Changing the subject line.)
> >
> > On Sat, 2008-10-18 at 05:54 -0500, Hyrum K. Wright wrote:
> >> Greg Stein wrote:
> >> > Should we yank 1.5.3 and publish a 1.5.4 right away?
> >> >
> >> > If this is a regression, then it seems we ought to pop out a 1.5.4 right away.
> >>
> >> Fine by me, if others feel that we need to. (Don't stay silent on this, people,
> >> I'd like a few +1's before going through the effort here.)
> >>
> >> -Hyrum
> >
> > >From the 1.5.x STATUS file:
> >> * r33693
> >> Fix merge bug introduced in the backport of the r29969, r32463, r33013,
> >> r33016, r33022, r33112 group in 1.5.3.
> >> Notes:
> >> There is a minor whitespace conflict in merge.c.
> >> Justification:
> >> This bug is a regression in 1.5.3, breaks mergeinfo aware merges when
> >> ecountered, and has already been seen a couple times in our own repos,
> >> see http://subversion.tigris.org/servlets/ReadMsg?list=dev&&msgNo=144187
> >> and http://subversion.tigris.org/issues/show_bug.cgi?id=3294#desc1.
> >> Votes:
> >> +1: pburba
> >>
--
Arfrever Frehtes Taifersar Arahesis
Re: Regression in merge - yank release 1.5.3? [was: svn commit: r33696 - branches/1.5.x]
Posted by Mark Phippard <ma...@gmail.com>.
I did not get the impression the bug was THAT bad. It sounds like you
are going by the wording of the backport proposal which makes it sound
like merge is completely broken. I am not against a 1.5.4 but I'd
like to hear from Paul. Perhaps we ought to just move up the 1.5.4
schedule so that it happens in early November?
Mark
On Sat, Oct 18, 2008 at 8:39 AM, Julian Foad <ju...@btopenworld.com> wrote:
> (Changing the subject line.)
>
> On Sat, 2008-10-18 at 05:54 -0500, Hyrum K. Wright wrote:
>> Greg Stein wrote:
>> > Should we yank 1.5.3 and publish a 1.5.4 right away?
>> >
>> > If this is a regression, then it seems we ought to pop out a 1.5.4 right away.
>>
>> Fine by me, if others feel that we need to. (Don't stay silent on this, people,
>> I'd like a few +1's before going through the effort here.)
>>
>> -Hyrum
>
> >From the 1.5.x STATUS file:
>> * r33693
>> Fix merge bug introduced in the backport of the r29969, r32463, r33013,
>> r33016, r33022, r33112 group in 1.5.3.
>> Notes:
>> There is a minor whitespace conflict in merge.c.
>> Justification:
>> This bug is a regression in 1.5.3, breaks mergeinfo aware merges when
>> ecountered, and has already been seen a couple times in our own repos,
>> see http://subversion.tigris.org/servlets/ReadMsg?list=dev&&msgNo=144187
>> and http://subversion.tigris.org/issues/show_bug.cgi?id=3294#desc1.
>> Votes:
>> +1: pburba
>>
--
Thanks
Mark Phippard
http://markphip.blogspot.com/
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org