[Mingw-users] FORW: Re: xs trouble on Win32 (building Win32::API)

[<so...@wo...>]

Found this <3a5957ad.2151874001@judy> in comp.lang.perl.modules:

== BEGIN forwarded message ==

From: ru...@sb... (Reini Urban)
Newsgroups: comp.lang.perl.modules
Subject: Re: xs trouble on Win32 (building Win32::API)
Date: Mon, 08 Jan 2001 06:54:30 GMT

Soren Andersen wrote:
>I am trying to compile the module Win32::API (http://dada.perl.it/) on WinNT4 
>sp6 using MinGW - gcc-2.95.2.1 and having a rough go of it with the xs code. 

>So, this _asm critter seems to be the first problem to tackle. Nobody that I 
>know of has built this mod using something else.
>Anybody (perhaps familiar with xs and Win32/MSVC++) got any insights?

inline _asm statements are not supported with egcs, so you will not be
lucky. you'll have to rewrite that particular function to the 
gcc supported _asm() syntax, which is different to MSVC.

I started porting the similar Win32::GUI to gcc, 
but this will need some time.

try this for the beginning, but I'm not sure how to do corect inline
assembler in GNUC.

abstr-cb.S:
.text
	.align 4
.globl _AbstractCallback
_AbstractCallback:
	popl  %eax
	ret

gcc -c abstr-cb.S
and link it to the others.


this fails for me:

void AbstractCallback() {
    int self;
#ifdef _MSC_VER
    _asm {
            pop  eax
            mov  dword ptr self, eax
    }
#else
    __asm__ ("popl %eax
              movl dword ptr self, %eax"
	     : "=ld" (self) /* output */
	     : 		   /* no input */
	     )
#endif
    printf("AbstractCallback: got eax=%ld\n", self);
}

-- 
Reini Urban
http://xarch.tu-graz.ac.at/autocad/news/faq/autolisp.html


== END forwarded message ==

I add this to the discussion. Thanks again to
all the great fellows who responded on this List.

I wish I could confidently say "Oh yeah, I am
going to tackle this one and make this port
work", but I cannot do so with any expectation
that it will happen soon. I simply don't have
an endless supply of spare time to work on it,
and believe me when I say it takes me ages to
accomplish even relatively small breakthroughs.

At least we've clearly identified the problem. What
would seem to me to be a great idea would be to have
a place where such porting challenges could be made
known to a cross-disciplinary population of progammers
(dada and I are basically Perl guys -- I for sure know
what my first love is ..) who could collaborate (as has
started happening very encouragingly on this List) on
such efforts. One Perl module doesn't seem like it is
really enough of a grabber to base an entire Sourceforge
project page around, for example, but some kind of central
node to which we would hope talented and ambitious
hackers would be drawn, might be a good step. I guess I
could really see this turning into a suggestion for a
Perl-specific section of MinGW project materials. Feedback?

    soren andersen