Re: [AD] get rid of ALLEGRO_MINGW32 and so on?

[Thomas Fjellstrom <tfjellstrom@xxxxxxxxxx>]

On May 31, 2010, Peter Wang wrote:
> On 2010-05-31, Thomas Fjellstrom <tfjellstrom@xxxxxxxxxx> wrote:
> > On May 31, 2010, Elias Pschernig wrote:
> > > Basically as I said in the 4.9.20 release thread on allegro.cc, it
> > > makes A5 distributions awkward if each differently built DLL needs a
> > > different set of headers even though the DLLs are compatible.
> > > 
> > > So, I think we should remove things like ALLEGRO_MINGW32 and use
> > > __MINGW32__ and so on instead which the compiler defines for us. Are
> > > there any apparent problems which would arise from that? If not I'll
> > > try make a patch for it.
> > 
> > I think the main thing it does is make allegro itself easier to
> > maintain. Inside its own code, you just use ALLEGRO_* defines, and not
> > the myriad of platform specific defines that may change between OS and
> > compiler versions.
> 
> I agree.
> 
> > Though I'm not entirely sure right now why they can't just all be
> > stuffed in a single header, and ship that single header instead of
> > several. That way all the code using ALLEGRO_* doesn't have to change,
> > and we still get a single header.
> 
> Not sure what you mean, they are all defined in alplatf.h right now.
> In some (all?) cases they could be defined like:
> 
> #ifdef __MINGW32__
>     #define ALLEGRO_MINGW32
> #endif
> 
> It would be nice to separate out the public and private headers.
> The public headers should be invariant across installations.

That is pretty much what I meant. Instead of bothering with a bunch of 
separate generated headers, we can just use one hard coded header with a 
bunch of ifdefs.

> Peter
> 
> -------------------------------------------------------------------------
> -----


-- 
Thomas Fjellstrom
tfjellstrom@xxxxxxxxxx