Re: Legacy applications and headers

[David Ayers]

Marcus Müller wrote:

> I'm currently evaluating a migration path from Apple's WebObjects 4.5 to 
> a SOPE/GDL2 based solution. When porting GDL2 to MacOSX Xcode I found 
> out that GDL2 often uses constructs like the following in its headers:
>
> #ifndef NeXT_Foundation_LIBRARY
> #include <Foundation/NSArray.h>
> #include <Foundation/NSException.h>
> #include <Foundation/NSZone.h>
> #else
> #include <Foundation/Foundation.h>
> #endif
>
> This is correct in a GNUstep environment, because some variables 
> concerning library combos and runtime are always defined there.
>
> Using Xcode on MacOSX  this is problematic because these variables are 
> not defined automatically. Sometimes it's possible to set these in the 
> respective build environments (Xcode, Makefiles), but this is a tedious 
> step and sometimes, for larger legacy projects, error prone - as I 
> currently experienced.
>
> (For completeness' sake, in the example above, without the manual 
> definition of NeXT_Foundation_LIBRARY the build will fail on MacOSX due 
> to redefinitions of NSArray and the like.)
>
>
> My request is to change these statements into:
>
> #ifdef GNUSTEP_BASE_LIBRARY
> #include <Foundation/NSArray.h>
> #include <Foundation/NSException.h>
> #include <Foundation/NSZone.h>
> #else
> #include <Foundation/Foundation.h>
> #endif
>
> It won't change anything in the GNUstep environment, however makes a bit 
> difference when using this on MacOSX.

I believe the test was intended to insure when GDL2 is build against 
other OPENSTEP-like Foundation libraries (like libFoundation maybe) that 
we only wanted to use the precompiled headers for Apple's Foundation, so 
therefore we need a test against some marker of that particular 
Foundation.  I was under the impression that NeXT_Foundation_LIBRARY was 
predefined by the Foundation headers, I guess I was wrong.  Maybe there 
is another marker we could use?

Now as Skyrix seems to have abandoned libFoundation (and I have no idea 
if GDL2 would have ever worked with it) we might have a true binary 
situation (ie. either GNUstep or Cocoa).  So the proposal may be a 
practical approach.  Yet, I'd like it to be consistent with the -core 
libraries, so if -core (ie. -baseadd) takes up this convention, -gdl2 
and -gsweb can follow.  (And as we include -baseadd headers you'll need 
the consistency also.)

Cheers,
David