Re: State of the pond

[David Kastrup]

Jean-Charles Malahieude <address@hidden> writes:

> Le 20/04/2012 23:42, Graham Percival disait :
>> On Fri, Apr 20, 2012 at 07:38:06PM +0200, Jean-Charles Malahieude wrote:
>>> Le 19/04/2012 21:30, Graham Percival disait :
>>>> - nobody touches the release/unstable branch, other than
>>>>    translators, who may merge with that if they want to and don't
>>>>    break anything.
>>>>    The question of whether translators have a stable branch or not
>>>>    is a separate matter and has nothing to do with the release
>>>>    plans.  It's just a question of how the translators want to
>>>>    organize themselves.
>>>
>>> Will you cherry-pick downloads from FTP, or should I push them
>>> directly on release/unstable?
>>
>> I'm not going to be cherry-picking downloads from FTP.  If you
>> think it's absolutely necessary to include those in the 2.16.0
>> release, and if you're not going to break anything, then push
>> directly to release/unstable.
>>
>> We do not have a long history of flawless git actions from
>> translations, so my preference would be not to touch anything.
>> That said, I can't recall any problems from FTP updates, so it's
>> probably ok.
>>
>
> Sorry, I didn't express myself very well.  I usually download updated
> po-files from FTP and push them to staging.
>
> It is important to have them with 2.16, since the "About automatic
> language selection" string seems to no longer be picked up from
> Documentation/po but from the binary's po. And some doc string, which
> I consider as important, have been modified.

I'd suggest that you do the same (namely push to staging), wait for
migration into master, check that the result agrees with your
expectations (I think the automated tests are not likely to cover this
well) and then ask Graham for permission to cherry-pick the respective
commit from master into release/unstable.  Test again and push.  And if
there is anything like mysterious merge conflicts or whatever else,
abort operations.  Regarding "do not have a long history of flawless git
actions from translations", most problems have been with automatic or
manual "salvaging" of mysterious conflicts.

-- 
David Kastrup