[FRC] Merge branch 'branch-1.11' into maint (was: Re: [maint] tests: put AM_PROG_CC_C_O before AC_OUTPUT to help losing compilers)

[Stefano Lattarini]

On 02/01/2012 10:33 PM, Peter Rosin wrote:
> Stefano Lattarini skrev 2012-02-01 22:31:
>> On 02/01/2012 09:31 PM, Stefano Lattarini wrote:
>>> On 02/01/2012 03:59 PM, Peter Rosin wrote:
>>>> When AM_PROG_CC_C_O is after AC_OUTPUT, the compile script
>>>> is not used even if needed, causing testsuite fails if
>>>> libtool is not used.
>>>>
>>>> * tests/depcomp8a.test: Uncomment the AM_PROG_CC_C_O macro
>>>> in its correct location, as indicated...
>>>> (configure.in): ...with this comment.
>>>> * tests/depcomp8b.test: Sync with tests/depcomp8a.test.
>>>>
>>> ACK.
>>>
>> Wait.  On a second thought, wait to push until I've complete
>> the unification between 'maint' and 'branch-1.11', OK?
>>
>> Thanks, and sorry for the confusion,
> 
> Too late.  Sorry.
> 
No problem luckily.  I hadn't really started the merge yet (I
had only written the commit message, which is below in case
you care to review it ;-) yet, so I'll just update my repo
before proceeding.

Thanks for the patience,
  Stefano

-*-*-

  Merge branch 'branch-1.11' into maint

  Keeping the 'branch-1.11' and 'maint' branches separated has become
  quite labour-intensive, and already caused too much confusion due
  to the entailed divergences (w.r.t. code for MSVC support) between
  the branches 'maint', 'master', 'branch-1.11' and 'msvc'.

  So we proceed to merge branch-1.11 into maint, and get rid of
  the msvc branch (since that was already routinely merged into
  branch-1.11).  This move shouldn't cause any regression in the
  stability of maint, as branch-1.11 was where we cut our maintenance
  releases from -- so it actually was even stabler and better tested
  than maint was.

  The only downside of this "unification" between branch-1.11 and
  maint will be an increased likelihood of merge conflicts taking
  place when we merge maint back into master.  But this can only
  happen in two scenarios:

   - right after an update to the Automake version number in
     'configure.ac' (which only happens when we cut a release
     from maint, so definitely not often); and

   - when we implement a *new* feature in both maint and master, but
     with the version in the two branches having slightly different
     behaviour between the; this has already happened, but only very
     seldom, and it's clear that it can only happen very seldom if a
     sane development process is maintained.