Re: [ANN] Lua-SDL2 rc1 available for testing!

[David Demelier <demelier.david@...>]

On 03/05/14 02:12, Rena wrote:
> On Fri, May 2, 2014 at 7:32 PM, Philipp Janda <siffiejoe@gmx.net
> <mailto:siffiejoe@gmx.net>> wrote:
>
>     Am 03.05.2014 00:40 schröbte Rena:
>
>
>         I just tried this and it wasn't able to find the library or
>         header files
>         for SDL2-ttf even after I installed libsdl-ttf2.0-dev.
>         I found the library in
>         /usr/lib/x86_64-linux-gnu/libSDL_ttf-2.0.so.0 and
>         symlinked it to /usr/local/lib/libSDL2_ttf.so but I wasn't
>         able to find
>         header files for it at all (only /usr/include/SDL/SDL_ttf.h
>         which I assume
>         is for SDL 1.x). I gave up there because I rather need text in my
>         application. I'm using Lubuntu 13.10 on AMD64 and trying to
>         compile for Lua
>         5.2.
>
>
>     The required package is `libsdl2-ttf-dev` which is in Ubuntu 14.04
>     but not 13.10. (I upgraded this weak precisely because of this).
>     If you don't want to upgrade just yet, you can try to install the
>     new packages manually via dpkg ...
>
>
>     Philipp
>
>
>
>
> Ah, I'll look into that. Not looking forward to upgrading though. (the
> auto upgrader has never worked for me, have to do a clean reinstall,
> and have been wanting to move away from Canonical stuff...)
>
> Also, found a bug in common/common.c, in function commonPush:
> commonPushUserdata(L, va_arg(ap, const char *), va_arg(ap, void *));
> This can result in the wrong pointers being given, because order of
> function argument evaluation in C is not defined. I ended up with
> renderers not having any metatable, because it was getting the wrong
> pointer for tname, not finding any such metatable, and thus setting
> nil as the metatable.
>
> I changed it to:
> case 'p': {
>     const char *name = va_arg(ap, const char *);
>     void *ptr = va_arg(ap, void *);
>     common_push_userdata(L, name, ptr);
>     ++ count;
>     break;
> }
> (the extra {} are needed to be able to declare variables within the
> scope of that case)
>
> Anywhere else you're using va_arg() more than once in a function call
> will similarly need to be changed.

Very interesting! I never known that calling a(b(), c()) will call b and
c in an undefined order. You taught me something thanks :-).

I already pushed the fix in the mercurial repository, thanks!

Regards,
David.