-mms-bitfields revisited

[Pau Garcia i Quiles]

2009/6/23 Bernhard Reiter <bernhard at intevation.de>:
> On Wednesday 17 June 2009, Patrick Spendrin wrote:
>> > It might be that current builds of KDE (without the right ABI) on windows
>> > interfacing with DLLs in the microsoft compiler layout will have somehow
>> > hidden defects because of this issue already.
>
> While this arguments still holds, it seems that our results have been
> obstructed by a mingw (/gcc) defect in some versions. Here is what Marcus
> Brinkman just found out last night for one struct used in gpgme gpgme++
> for offsetoffs:
>   mingw 4.2.1 (Ubuntu 9.04 == Debian Lenny)
>   gcc                 0x20
>   gcc -mms-bitfields  0x20
>   g++                 0x20
>   g++ -mms-bitfields  0x20
>
>   mingw 3.4.5 (Debian Etch)
>   gcc                 0x20
>   gcc -mms-bitfields  0x30 !!!
>   g++                 0x20
>   g++ -mms-bitfields  0x30 !!!
>
> So there seems be a gcc/mingw defect involved as well.
> Still in other cases we would need the right ABI for the platform
> so I guess using ms-bitfields would still be correct.

mingw-gcc 4.4.0, with good exception support, has been released today
as a replacement of mingw-gcc 3.4.5. I think that's the version KDE on
Windows-MinGW  should be based from now on.

-- 
Pau Garcia i Quiles
http://www.elpauer.org
(Due to my workload, I may need 10 days to answer)