Re: long_long and ulong_long deprecated

[stefan]

On 6 Sep 2001, Marius Vollmer wrote:

> > The exact problem is about "long long" is MSVC's causing a parse error. It
> > does not know it. But it knows "__int64" which is the current equivalent
> > to gcc's "long long".
> 
> Hmm, why does it encounter long long at all?  All occurrences should be
> guarded by HAVE_LONG_LONG, and this should not be defined on your
> platform.

I defined it in my config.cache by hand, because I knew that there is a
"long long" replacement (which is __int64). Thus I have a HAVE_LONG_LONG
in my <config.h>.

> Can you give more details?  Like, whether configure thinks you have
> long longs, and where exactly the compiler fails because of them?

Configure itself is yet unable to detect "long long" support. In my
previous port of Guile 1.4 I used the ulong_long and long_long typedef
(which I therefore considered a very good thing) and redefined them.

You can see these changes in the CVS diff I sent to you and Martin
Grabmueller before.

I can resend the complete diff any time again... It enables the M$ VisualC
and the MinGW gcc to build the libguile on Windows.

Of course there is yet pending things I would like to see in Guile making
it easier to port to Windows:

 * the long discussed SCM_API prefix

This would enable me to make a complete port with all features (readline,
crypt, srfi, etc.). You can prove this trying the native port of Guile 1.4
of myself.

I know some people "do not care about Windows for years" on this list, but
there are some projects and people I know who do. That is why I keep
trying...

Cheers,
        address@hidden