[svn.haxx.se] · SVN Dev · SVN Users · SVN Org · TSVN Dev · TSVN Users · Subclipse Dev · Subclipse Users · this month's index

Re: 1.5.x branch serf tests are failing

From: David Glasser <glasser_at_davidglasser.net>
Date: Sun, 25 May 2008 11:20:23 -0700

On Sun, May 25, 2008 at 2:08 AM, Lieven Govaerts <svnlgo_at_mobsol.be> wrote:
> Mark Phippard wrote:
>>
>> On Sat, May 24, 2008 at 2:21 PM, Mark Phippard <markphip_at_gmail.com> wrote:
>>>
>>> On Sat, May 24, 2008 at 12:45 PM, Mark Phippard <markphip_at_gmail.com>
>>> wrote:
>>>>
>>>> On Sat, May 24, 2008 at 12:27 PM, Mark Phippard <markphip_at_gmail.com>
>>>> wrote:
>>>>>
>>>>> Can someone try running the Serf tests with the current branch? I
>>>>> have tried extraclean and several other things but tests are failing.
>>>>> I think it is caused by this backport which we may have to
>>>>> reverse-merge from the branch.
>>>>
>>>> The tests seem to be passing for Hyrum, so my build environment might
>>>> be messed up. I am going to re-configured using the Serf 0.1.2 from
>>>> MacPorts.
>>>>
>>>> Neon tests seem to be fine. So hopefully this is a false alarm.
>>>
>>> I have done everything I can think of and I still get lots of Serf
>>> test failures on 1.5.x branch. I removed my 1.5 WC and checked out
>>> again. I removed all traces of Serf and reinstalled it from MacPorts.
>>> Same errors as I was originally getting when using serf-trunk.
>>>
>>> Can we get some more people to try these tests? I think we might need
>>> to reverse-merge r31357 so that we can make the release. It does not
>>> look critical anyway and can go into 1.5.1 if we figure out what is
>>> wrong.
>>>
>>> My serf tests seemed to be passing in trunk. I did not run them all
>>> the way, just past some of the places I am getting errors in 1.5.x.
>>> So maybe there are some other trunk changes that needed to go with
>>> this one?
>>
>> I wound up reverse-merging this change so that we could do the
>> release. I ran all tests and they all passed with this change
>> removed. The change did not seem critical for the release and could
>> just be included in 1.5.1 once we sort out why the tests were failing
>> for me.
>>
>
> Good call Mark.
>
> I found that this change relies on r31234 which was not merged to 1.5.x.
> Besides that, in some situations it's leaking an error causing an abort in
> debug mode.
>
> With 1.5 ready to be released, and us not making API changes in patch
> releaes this will be added in 1.6.

We should probably get r31234 into 1.5.1 anyway.

I think it would be fine to put r31357 in 1.5.1 too. Sure, it's an
"API change", but it's the sort of API change that, if we consider it
anything other than a bugfix, would be illegal under our rules before
2.0 anyway...

--dave

-- 
David Glasser | glasser@davidglasser.net | http://www.davidglasser.net/
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe_at_subversion.tigris.org
For additional commands, e-mail: dev-help_at_subversion.tigris.org
Received on 2008-05-25 20:20:43 CEST

This is an archived mail posted to the Subversion Dev mailing list.