<p>Am 19.08.2016 10:56 schrieb "OBones" <<a href="mailto:obones@free.fr">obones@free.fr</a>>:<br>
><br>
> Sven Barth wrote:<br>
><br>
>><br>
>> Am 19.08.2016 09:55 schrieb "Jonas Maebe" <<a href="mailto:jonas.maebe@elis.ugent.be">jonas.maebe@elis.ugent.be</a> <mailto:<a href="mailto:jonas.maebe@elis.ugent.be">jonas.maebe@elis.ugent.be</a>>>:<br>
>> > As alluded to above, LLVM support needs to be added/tested/maintained<br>
>> > separately for each supported architecture and to a lesser extent for<br>
>> > each supported OS. Right now, I only have plans for x86-64 and AArch64<br>
>> > (and maybe PowerPC64), on Darwin and Linux. Personally, I won't add<br>
>> > support for Windows since I don't use that platform. Support for 32 bit<br>
>> > platforms in general will be a bit tricky due to the way our compiler is<br>
>> > structured.<br>
>><br>
>> Why is it that 32-bit targets would be a bit tricky to implement?<br>
>><br>
>> Regards,<br>
>> Sven<br>
>><br>
> I believe because of this:<br>
><br>
><br>
> There are also a few LLVM limitations over which I have no influence:<br>
> <snip><br>
><br>
> b) LLVM has no support for the i386 "register" calling convention, so I<br>
> will probably never add support for the i386 target using LLVM</p>
<p>Jonas wrote about the structure of the compiler. "register" support is not about the structure, it's merely an additional calling convention.</p>
<p>Regards,<br>
Sven</p>