Thought about a Qt port

[Lubomir I. Ivanov]

On 30 March 2013 20:37, Thiago Macieira <thiago at macieira.org> wrote:
> On sábado, 30 de março de 2013 10.40.50, Dirk Hohndel wrote:
>> > For Windows, I'd recommend building yourself. The current binary downloads
>> > are for hosted builds, plus the MinGW folks can't make up their minds
>> > about what ABI to use.
>>
>> Bleh. Not having a stable MinGW build really hurts me. Maybe we can
>> engage with them while we do this and help figure this out?
>
> Trust me, I have. I have knocked heads on this problem before and I've even
> said anyone still using SJLJ exceptions should be taken out and shot (Linus
> would've called them morons). And lo and behold! the Qt 5.0.1 MinGW binaries
> we produced used SJLJ and within a week we got bug reports of extreme
> performance regressions.
>
> I won't bother you with details. This is being taken care of and should much
> improve with GCC 4.8 released.
>
> Moreover, it does not affect Subsurface. Whichever exception ABI is chosen, it
> will work with Qt and the Windows installation will bring in all the necessary
> libraries anyway.
>
> PS: I think the Fedora cross-mingw build uses SJLJ → performance problems.
>

out of curiosity - are you shooting for a Mingw with SEH (if that even
works yet in 4.8 with the TIBs and the handler linked lists and such)
or DW2 to be distributed with the next (perhaps 5.1 minor) release?
though, if DW2 is chosen i remember there were some issues with that
and how the WINAPI works, but i could be wrong.

also, there were some nice improvements in ABI compatibility with MSVC
in 4.8 (or was it back in 4.7), but a combo of C++ virtual functions
and some varargs (cdecl x86) really got the best out of it, so i gave
up testing further at that point. i think it was early 2012...

lubomir
--