Re: For Bug Squad (was Re: Bad translation merge)

[David Kastrup]

James <address@hidden> writes:

> On 8 March 2012 00:20, David Kastrup <address@hidden> wrote:
> ...
>
>>
>> I have checked the versions, and it would appear that once translation
>> has been checked _and_ the new master merged into it _and_ back into
>> staging again, the Bug Squad will need to reverify all issues with
>> Fixed_2_15_31 or later _textually_ (meaning that they need to check that
>> the changes are in the work tree, not that the commit is in the history:
>> the commits will be there several times under different commit ids, and
>> some of the commit history is a lie).
>
> If it helps you, I have made this tiny url that will (hopefully) list
> all the 'Verfied' issues for Fixed_2.15.31 and Fixed_2.51.32
>
> I get about 30 issues here.
>
> http://tinyurl.com/7xan8v3
>
> @David/Graham, when David says '... the changes are in the work tree,
> not that the commit is in the history', I am not getting the subtlety
> here. I assume we can no longer, for these commits, trust what we see
> in
>
> http://git.savannah.gnu.org/gitweb/?p=lilypond.git

Yes.  You have to look at

http://git.savannah.gnu.org/gitweb/?p=lilypond.git;a=tree

instead.  I apologize for the inconvenience, but the problematic commit
introduced a history that claimed to contain a number of patches which
it didn't.  And my fix did not remove that problematic history, but
introduced an artificial sideline supposed to cancel its effects.

It is probably sufficient to look at each of the claimed commit ids
fixing the problem, and try _reverting_ the commit on top of current
master (of course, not pushing the result but rather resetting one's
private master afterwards each time).

If that
a) has an effect
b) does not complain that the reverse patch does not apply

it is likely that the effect of the commit survived into master.

-- 
David Kastrup