couchdb-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From J Chris Anderson <jch...@gmail.com>
Subject Re: 1.0 Vote
Date Fri, 25 Jun 2010 17:46:12 GMT

On Jun 25, 2010, at 10:36 AM, Jan Lehnardt wrote:

> Chris,
> 
> we've been bending the rules already by committing code to 0.11.x that clearly isn't
just bugfixes. If we are going to to cut 1.0 from trunk, I'd ask you to identify and revert
all the backports that don't fit that bill from 0.11.x.
> 

I don't mind doing legwork to get this right. What matters is getting good code. I'm ready
to pull the trigger on an all-day rebase of trunk into 0.11.x, but I want to know which patches
should be excluded. If the list is short enough then I'll be happy to branch trunk and then
remove the 1.1-only patches.

Sorry if this means backport work done in the past was unnecessary, but that's not as important
as a stable clean 1.0

> I don't care much either way about where to cut 1.0 from at this point except I wish
this would have been brought up earlier so I didn't had to waste a lot of time on the backports.
> 

I've been skeptical of the "branch 1.0 from 0.11.x plan" the entire time, because I know how
stuff goes. There are lots of tiny fixes that get dropped into trunk and backported or not,
and then, when it comes release time, we read the trunk code and realize that it should pretty
much all go in the release.

A major concern of mine is that in our attempts to be conservative in the push from 0.11 to
1.0, we don't end up with a 1.0 that is a backporty mess, which makes maintenance between
1.0, 1.1, 1.2 etc a big pain because there are lots of eg whitespace patches (or method renames)
etc that are in 1.1 but not 1.0.

Chris

> Cheers
> Jan
> --
> 
> 
> 
> On 25 Jun 2010, at 19:23, J Chris Anderson wrote:
> 
>> I'm a little confused.
>> 
>> Last week I mentioned that we should cut trunk from 1.0, after reviewing the diff
between 0.11.x and trunk, and seeing that every change in trunk I could be confident is release
quality.
>> 
>> Now we've done a bunch of backporting to 0.11.x, which is bringing it closer to trunk,
but not close enough, imho.
>> 
>> We could continue to backport to 0.11.x until `git diff trunk 0.11.x` is empty, but
that seems like crazy lots of work to no good end.
>> 
>> Are there any patches in trunk which do not belong in 1.0?
>> 
>> Jan mentioned that Adam has some writer patches that he wants in 1.1. Is there anything
else?
>> 
>> I would *definitely* like to backport the couch_util:get_value patch. There's no
reason to have that huge diff between 1.0 and 1.1 (that doesn't change functionality). The
big diff just makes it harder to maintain 1.0 going forward.
>> 
>> I don't care which route we get there by, but at this point, there is lot of code
in trunk that belongs in 1.0
>> 
>> It might be simpler to branch trunk, and remove the one or two commits that shouldn't
be in 1.0, than to play the rebase game even longer, trying to get 0.11.x to look like trunk.
>> 
>> Can you help me identify the commits that should *not* be in 1.0? 
>> 
>> Thanks!
>> 
>> Chris
>> 
>> 
>> On Jun 25, 2010, at 8:35 AM, Filipe David Manana wrote:
>> 
>>> On my GNU/Linux machine, branch 0.11.x, I have the test list_views.js with
>>> an assertion failing:
>>> 
>>> Assertion failed: expected '{"total_rows":10,"offset":0,"update_seq":11}',
>>> got '{"total_rows":10,"offset":0}
>>> 
>>> 
>>> 
>>> On Fri, Jun 25, 2010 at 1:30 PM, Noah Slater <nslater@tumbolia.org> wrote:
>>> 
>>>> Where are we on Windows support?
>>>> 
>>>> The last time I heard, the unofficial Windows binaries that can be built
at
>>>> the moment are not ready for the prime time. Is this still the case. If
>>>> we're doing a windows release, who is going to prepare the release artefact?
>>>> 
>>>> On 25 Jun 2010, at 01:20, Damien Katz wrote:
>>>> 
>>>>> It looks like the file deletion issue may be already fixed windows. If
we
>>>> can get confirmation the Windows compaction and deletion problems are fixed,
>>>> I think we can prepare a release artifact to vote on.
>>>>> 
>>>>> -Damien
>>>>> 
>>>>> 
>>>>> On Jun 24, 2010, at 12:24 PM, Jan Lehnardt wrote:
>>>>> 
>>>>>> 
>>>>>> On 24 Jun 2010, at 01:22, Damien Katz wrote:
>>>>>> 
>>>>>>> 
>>>>>>> On Jun 23, 2010, at 4:15 PM, Paul Davis wrote:
>>>>>>> 
>>>>>>>> On Wed, Jun 23, 2010 at 6:47 PM, Damien Katz <damien@apache.org>
>>>> wrote:
>>>>>>>>> I think we are really close to being ready for 1.0.
>>>>>>>>> 
>>>>>>>>> Are there any issues/patches that need to be handled
before we make a
>>>> release and vote on it?
>>>>>>>>> 
>>>>>>>>> Noah, do you have any issues, and can you make the branch
and release
>>>> candidate?
>>>>>>>>> 
>>>>>>>>> One thing I can think of is windows file support, so
that compaction
>>>> and deletion work properly. There exists a fix for it from Mark Hammond,
but
>>>> we need a patched Erlang VM before the fix can work. The upcoming R14B
>>>> includes the VM patch, but I think we should wait until it's actually
>>>> released before we include Mark's fix.
>>>>>>>>> 
>>>>>>>>> -Damien
>>>>>>>> 
>>>>>>>> If you mean the SHARED_DELETE thing for opening files, that
was
>>>>>>>> included in R14A which was released the other day.
>>>>>>> 
>>>>>>> I do mean the SHARED_DELETE thing, and I didn't know that. Great
news!
>>>>>>> 
>>>>>>> Anyone up for merging the patch and testing on windows?
>>>>>>> 
>>>>>>> Patch is here: https://issues.apache.org/jira/browse/COUCHDB-86
>>>>>> 
>>>>>> The patch is stale, we'd need somebody to get it applying to trunk.
>>>> Anyone?
>>>>>> 
>>>>>> Cheers
>>>>>> Jan
>>>>>> --
>>>>>> 
>>>>> 
>>>> 
>>>> 
>>> 
>>> 
>>> -- 
>>> Filipe David Manana,
>>> fdmanana@apache.org / fdmanana@gmail.com
>>> 
>>> "Reasonable men adapt themselves to the world.
>>> Unreasonable men adapt the world to themselves.
>>> That's why all progress depends on unreasonable men."
>> 
> 


Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message