Re: [Mingw-w64-public] Latest mingw64 trunk failed to build gcc4.6.2's libgfortran

[Kai T. <kti...@go...>]

2011/11/13 Ozkan Sezer <se...@gm...>:
> On Sun, Nov 13, 2011 at 12:27 PM, Kai Tietz <kti...@go...> wrote:
>> 2011/11/13 Ozkan Sezer <se...@gm...>:
>>> [...]
>>>> Ozkan,  I would like to see the changes done to stdio.h, wchar.h (and
>>>> related _mingw_print_(push/pop).h on stable v2 branch, too.  They are
>>>> fixing an issue we have with libstdc++, as without this patch
>>>> input/output stream of 'long double' types won't work.
>>>>
>>>> Regards,
>>>> Kai
>>>
>>> Done:  Please make sure that I didn't screw-up anything.
>>>
>>> Regarding backports:
>>>
>>> -  Do we want the _mingw_print_push.h, _mingw_print_pop.h, stdio.h and
>>>  wchar.h changes in v1.x?
>>
>> Well, I would say no.  But of course it is up to you.  It fixes an
>> issue in for libstdc++, so it might be seen as bugfix.
>>
>>> -  Do we want r4591 in v2.x and/or v1.x?
>>
>> Well, for v2 I think it make sense.
>
> OK
>
>>   For v1.x not sure.
>>   Btw how long
>> we want to keep branch v1.x maintained?
>
> I'm merging obvious + bugfixes-only.  For how long, well, until I get
> bored (and I am beginning to :)
>
> So, I'd like an admin decision: are the stdio and/or *.def/assert
> thingies eligible for v1?
>
>>
>> Regards,
>> Kai
>
> --
> O.S.

So, the *.def changes for _assert/_wassert are somthing important for
v2.0 and for v1.0.  The avoid a multiple definition link failure in
some cases.

The stdio changes are important for working 'long double'
stream-input/output for libstdc++.  So I see it as important for v2.x,
and dependent to live-time of v1 also for it.

Kai