[Releasing] Meeting minutes: Qt release team meeting 12.01.2016

[Knoll Lars]

On 15/01/16 01:25, "Releasing on behalf of Thiago Macieira" <releasing-bounces at qt-project.org on behalf of thiago.macieira at intel.com> wrote:



>On Thursday 14 January 2016 18:24:46 Knoll Lars wrote:
>> >I'd rather apply the fix instead.
>> 
>> 
>> Well, this works in 5.6, I guess because we're not using std::atomic there.
>> Why aren't we continuing to use that code path until QNX has released a
>> fixed tool chain?
> 
>Because it's the old atomic code, some of which I rewrote for 5.0, most of 
>which are the same ones Brad made public for 4.4, with the initial 
>implementations from 4.0.
>
>There are problems with them, as evidenced by the bug reports received in the 
>last year. Some are serious problems, like the 32-bit ARM code for 64-bit 
>integers. And I think that the AArch64 codepath is just completely busted -- 
>unless Q_COMPILER_ATOMICS is defined, in which case we are back where we 
>started.

Well, I wouldn’t have any problems if we in this case disable atomics for 64bit integers on 32bit (I don’t think we use those anywhere ourselves), and require proper atomics for AARCH64. I think we only support 32bit on QNX anyway.

And of course switch over to C++11 atomics as soon as a fixed toolchain is out.

Cheers,
Lars

>
>I don't test non-IA and, most importantly, I don't want to maintain the hand-
>written assembly anymore.
>
>-- 
>Thiago Macieira - thiago.macieira (AT) intel.com
>  Software Architect - Intel Open Source Technology Center
>
>_______________________________________________
>Releasing mailing list
>Releasing at qt-project.org
>http://lists.qt-project.org/mailman/listinfo/releasing