Re: Legacy applications and headers

[Helge Hess]

Hi Nicola,

some comments below. As mentioned I don't really care about the way 
this is done, so please feel free to ignore them instead of making a 
huge thread out of this minor issue ;-)
(personally I would recommend to use only one of the approaches, I find 
clean headers much more important than a 1s faster compile).

On Mar 19, 2004, at 5:51 PM, Nicola Pero wrote:
> I slightly disagree :-) ... This is a speed optimization for GNUstep 
> :-)

Only if PCH is considered the default case ;-)

> I think the main concept is:
...
> So I think GNUSTEP is a nice define to use in this case.

Well besides the point that the GNUSTEP define only signals a certain 
library combo and has semantically nothing to do with precompiled 
headers, as mentioned you are going to rewrite all the headers once PCH 
is available for GNUstep.
This really doesn't make a lot of sense to me.

> And yes you have a point with gnustep-base getting a PCH compiler, but 
> I'd ignore this complication for now.

Which is why I suggested the __APPLE__ define. David is right that it 
slows down gnu-gnu-gnu compilation on MacOSX but it is perfect for any 
other case in the moment and I guess gnu-gnu-gnu compilation on OSX 
isn't a very common one ;-)

> But you reminded me that I need to look at the new PCH support in GCC, 
> and  on how we have gnustep-make automatically take advantage of that!

:-)

Greets,
  Helge
--
http://docs.opengroupware.org/Members/helge/
OpenGroupware.org