On Sun, 6 Jan 2013, Zoltán Herczeg wrote:
>> Everything builds and tests fine as-is. I removed the push/pop that had
>> been added in sljitNativeX86_common.c, and that builds/tests fine too.
>
> Thank you for trying it. The question is shall we make the ebx save GCC
> specific or not. At the moment I would say not, as it seems it does not
> hurt other compilers, but I don't have any strong opinion. What do you
> think? Or anyone else on the list?
Is it safer to do the push/pop (so ebx isn't clobbered), or not to do the
push/pop (so the stack isn't touched)? I couldn't tell from the comment in
the code.
Presuming it's safer to push/pop, I would do that regardless of the
compiler. It's simpler, frees us from worrying about compiler smarts, and
I'm guessing this isn't performance-critical code anyway.
--Daniel
--
Daniel Richard G. || danielg@??? || Software Developer
Teragram Linguistic Technologies (a division of SAS)
http://www.teragram.com/
