gamedevlists-general

Re: [GD-General] Clean way of generalizing CDECL

[brian s. <pud...@po...>]

I never came up with anything better than CDECL_PRE and CDECL_POST but
that solution sure won't win any beauty contests :).

What did the MingW people do for this?  All of the Windows API
functions are stdcall, and surely they didn't modify every Windows
header to change the declarations.  They must be able parse them
as-is.  Maybe there's a solution to be found there?

--brian

> I'd like to be able specify CDECL in a cross-platform manner.  On most
> x86 compilers there's just the __cdecl keywod:
>
> void __cdecl foo( int x, int y, int z );
>
> GCC, of course, does it completely differently:
>
> void foo( int x, int y, int z ) __attribute((cdecl));
>
> So a simple #define won't work.  Now, you can do a kind of convoluted
> thing using a define, but damn, it's REAL ugly.
>
> Is there a cleaner way of doing this?
>

RE: [GD-General] Clean way of generalizing CDECL

[Nick T. <ni...@ro...>]

> -----Original Message-----
> From: gam...@li...
> [mailto:gam...@li...] On Behalf Of
> Brian Hook
> Sent: Tuesday, July 06, 2004 3:11 PM
> To: gam...@li...
> Subject: [GD-General] Clean way of generalizing CDECL
>=20
> I'd like to be able specify CDECL in a cross-platform manner.  On most
> x86 compilers there's just the __cdecl keywod:
>=20
> void __cdecl foo( int x, int y, int z );
>=20
> GCC, of course, does it completely differently:
>=20
> void foo( int x, int y, int z ) __attribute((cdecl));
>=20
> So a simple #define won't work.  Now, you can do a kind of convoluted
> thing using a define, but damn, it's REAL ugly.

Why not:

#define foo(T,N, ...) T __cdecl N ( __VA_ARGS__ )

foo (int, blah, int a, int b)  // declaration

or

#define foo2(T,N,A) T __cdecl N A=20

foo2 (int, blah, (int a, int b))  // declaration


Nick

RE: [GD-General] Clean way of generalizing CDECL

[Brian H. <ho...@bo...>]

> Why not:
>
> #define foo(T,N, ...) T __cdecl N ( __VA_ARGS__ )

__VA_ARGS__ is C99, which isn't widely supported yet.

> #define foo2(T,N,A) T __cdecl N A
>
> foo2 (int, blah, (int a, int b))  // declaration

Nick, LOOK at that line for a second, and then ask yourself why I 
wouldn't want to write something like that =3D)

Brian

RE: [GD-General] Clean way of generalizing CDECL

[Nick T. <ni...@ro...>]

> > foo2 (int, blah, (int a, int b))  // declaration
>=20
> Nick, LOOK at that line for a second, and then ask yourself why I
> wouldn't want to write something like that =3D)

I think that looks like an excellent Lisp example. :) It works doesn't
it! I'd rather see that than loads of #ifdefs (e.g. see GLUT headers)
And you can easily add new platforms which also don't comply later.=20

Nick

RE: [GD-General] Clean way of generalizing CDECL

[Stefan B. <Ste...@di...>]

  Why do you want to generalize it? Isn't cdecl something very
win32-specific anyway?

/Stefan

-----Original Message-----
From: gam...@li...
[mailto:gam...@li...] On Behalf Of
Brian Hook
Sent: 07 July 2004 00:11
To: gam...@li...
Subject: [GD-General] Clean way of generalizing CDECL

I'd like to be able specify CDECL in a cross-platform manner.  On most=20
x86 compilers there's just the __cdecl keywod:

void __cdecl foo( int x, int y, int z );

GCC, of course, does it completely differently:

void foo( int x, int y, int z ) __attribute((cdecl));

So a simple #define won't work.  Now, you can do a kind of convoluted=20
thing using a define, but damn, it's REAL ugly.

Is there a cleaner way of doing this?




-------------------------------------------------------
This SF.Net email sponsored by Black Hat Briefings & Training.
Attend Black Hat Briefings & Training, Las Vegas July 24-29 -=20
digital self defense, top technical experts, no vendor pitches,=20
unmatched networking opportunities. Visit www.blackhat.com
_______________________________________________
Gamedevlists-general mailing list
Gam...@li...
https://lists.sourceforge.net/lists/listinfo/gamedevlists-general
Archives:
http://sourceforge.net/mailarchive/forum.php?forum_idU7

RE: [GD-General] Clean way of generalizing CDECL

[Brian H. <ho...@bo...>]

> Why do you want to generalize it? Isn't cdecl something very win32-
> specific anyway?

Yes and no.  cdecl, fastcall, stdcall, etc. are x86 specific (not just 
Win32), but many platforms still have multiple calling conventions 
(e.g. MacOS on the 68K had at least three if I remember: pascal, OS, 
and something else).  So you still need to be able to explicitly 
specify calling conventions somehow in a vaguely cross-platform 
manner.

Brian

RE: [GD-General] Clean way of generalizing CDECL

[Ivan-Assen I. <as...@ha...>]

> Yes and no.  cdecl, fastcall, stdcall, etc. are x86 specific 
> (not just Win32), but many platforms still have multiple 
> calling conventions (e.g. MacOS on the 68K had at least three 
> if I remember: pascal, OS, and something else).  So you still 
> need to be able to explicitly specify calling conventions 
> somehow in a vaguely cross-platform manner.

Ummm, ok, let's rephrase this: WHY do you need to explicitly specify
calling conventions in a cross-platform manner? It's not like you'd be
linking object files from one platform against the libs of the other.

RE: [GD-General] Clean way of generalizing CDECL

[Brian H. <ho...@bo...>]

> Ummm, ok, let's rephrase this: WHY do you need to explicitly
> specify calling conventions in a cross-platform manner? 

I don't, but I need to be able to have them specified on one platform 
and then not have them fail when compiled on another platform.

Brian

Re: [GD-General] Clean way of generalizing CDECL

[Javier A. <ja...@py...>]

Ivan-Assen Ivanov wrote:

> Ummm, ok, let's rephrase this: WHY do you need to explicitly specify
> calling conventions in a cross-platform manner? It's not like you'd be
> linking object files from one platform against the libs of the other.

Most likely because you want to share many interfaces (header files) across
platforms, therefore you want declarations to include "some" calling
convention. Platform-specific configuration headers would #define the
platform-specific details of that calling convention. It's not too different
from the common technique used in Win32 DLL declarations:

// DLLFunctions.h
#ifdef DO_DLL_EXPORT
#define DLL_FUNCTION __declspec(dllexport)
#else
#define DLL_FUNCTION __declspec(dllimport)
#endif

extern "C" DLL_FUNCTION int DLLFunction1();
extern "C" DLL_FUNCTION void DLLFunction2();

// DLLFunctions.cpp
#define DO_DLL_EXPORT
#include "DLLFunctions.h"

// ...DLL functions implemented here

// DLLUser.h
#include "DLLFunctions.h"

// ...Code here uses the DLL functions

The original problem was that, since different platforms specify calling
conventions in different places (some before the function, some after the
parameters, etc), such a cross-platform call convention specifier may end up
being too ugly. I don't think there are much better ways than suggested by
Brian Sharon and Nick Trout. Techniques similar to Nick's are widely used in
K&R-conforming portable libraries like ZLib:

ZEXTERN int ZEXPORT inflate OF((z_streamp strm, int flush));

The OF macro here is used to remove the parameters from declarations in K&R
platforms.

-- 
  Javier Arevalo
  Pyro Studios

Re: [GD-General] Clean way of generalizing CDECL

[Brian H. <ho...@bo...>]

> Most likely because you want to share many interfaces (header
> files) across platforms, therefore you want declarations to include
> "some" calling convention. 

Exactly.  

And it's necessary to specify calling conventions so that your library 
compiled with one convention is compatible with someone else's 
application that may have been built with __fastcall.

PoshLib (www.poshlib.org) tries to solve most of these problems 
cleanly and portably, which is why I'm asking about this.  I'm glad to 
find out that __attribute__((cdecl)) seems to be prefix or suffix 
friendly.

With POSH you do:

void POSH_CDECL foo( int x );

or, if you ant DLL export to work magically as well:

POSH_PUBLIC_API( void ) POSH_CDECL foo( int x );

Brian

Re: [GD-General] Clean way of generalizing CDECL

[Jeremy V. <mad...@gm...>]

Hey guys,

The way I would choose, which is easy on the eyes, is to always define
__cdecl functions with both, and then have defines that eliminate one
syntax or the other depending on the platform. Such as:

#ifdef WIN32
# define __attribute(x) 
#else
# define __cdecl 
#endif

int __cdecl myFunction(double whatever) __attribute((cdecl));

Actually, this is the same as Brian's CDECL_PRE and CDECL_POST - so I
should probably just stick to lurking :p

[GD-General] Free downloadable isometric tilesets?

[Brian H. <ho...@bo...>]

I'm trying to find some tiles to prototype a simple (2D) isometric 
engine, and unlike sprites or 3D models it seems that there's a tough 
time finding iso (2:1, diamond map) art work out there.

Anyone know of a good resource (cheap commercial or free) suitable for 
this?

Thanks,

Brian

Re: [GD-General] Free downloadable isometric tilesets?

[Peter B. <fla...@gm...>]

there's falcon's eye the isometric nethack (not sure what the 'nethack
public license' is tho').

Shots:
http://www.hut.fi/~jtpelto2/nhfeatures.html

dowload:
http://falconseye.sourceforge.net/

On Mon, 2 Aug 2004 20:18:53 -0400, Brian Hook <ho...@bo...> wrote:
> I'm trying to find some tiles to prototype a simple (2D) isometric
> engine, and unlike sprites or 3D models it seems that there's a tough
> time finding iso (2:1, diamond map) art work out there.
> 
> Anyone know of a good resource (cheap commercial or free) suitable for
> this?
> 
> Thanks,
> 
> Brian
> 
> -------------------------------------------------------
> This SF.Net email is sponsored by OSTG. Have you noticed the changes on
> Linux.com, ITManagersJournal and NewsForge in the past few weeks? Now,
> one more big change to announce. We are now OSTG- Open Source Technology
> Group. Come see the changes on the new OSTG site. www.ostg.com
> _______________________________________________
> Gamedevlists-general mailing list
> Gam...@li...
> https://lists.sourceforge.net/lists/listinfo/gamedevlists-general
> Archives:
> http://sourceforge.net/mailarchive/forum.php?forum_idU7
>