Re: SV: [xml] windows binary with different calling

[Igor Zlatkovic <igor zlatkovic com>]

> >  xmlDocPtr XMLCALL xmlParseFile(xmlChar* filename);
> >
> > or even this:
> >
> >  XMLEXPORT xmlDocPtr XMLCALL xmlParseFile(xmlChar* filename);
>
>
>   I think that's the format, of course currently all exports are formatted
> in the header files to align arguments, this will be a pain... The name of
> the macro is not binding but 1/ XML prefix and 2/ short to minimize space
> are 2 constraints I would like to impose ... XMLEXPORT/XMLCALL are fine.
> Check withn Mark for this ...

The names of those macros bear no importance to me, any is as good as 
any other. I agree about the XML prefix and them being as short as they 
can get.

I suggest we do it with one of the headers and post the changed header. 
Then you can take a good look at it and see how you feel about the taint :-)

> > Once that is clear, and the scripts are settled, some of us others will do
> > the work, it is our platform. :-) Really, no point in you doing things you
> > cannot test, or have you installed Windows in the meantime? :-)
>
>
>   Nahhh ! No Way !!!
> But I will have to make sure the macros really get defined to empty strings
> when building on anything but Windows (and apparently OS/2)

When we do a

  #if defined _WIN32
    #define XMLCALL __cdecl
  #else
    #define XMLCALL
  #endif

then I am quite sure every Unix incarnation will have it empty... unless 
you compile with -D_WIN32 on your box :-) I think it safe to assume the 
sun will go nova before you do so :-) :-)

Ciao,
Igor