> -----Original Message-----
> From: Prof Brian D Ripley [SMTP:ripley@stats.ox.ac.uk]
> Sent: Monday, January 25, 1999 2:48 PM
> To: Clark Sims
> Subject: Re: [S] [Fwd: Mixed language programing with SPLUS 4.5
and
> WATCOM C++ 10.5-11.0]
>
> On Mon, 25 Jan 1999, Clark Sims wrote:
>
> That was a very long message! The comments about register vs
stack-based
> _are_ critical for dyn.load, which you do not seem to mention,
strangely,
> as it is by far the easiest way to use Watcom compilers, and is what
the
> COMPILE script was designed for. As I have ported most of the code on
> Statlib that way I am rather sure is works reliably on a wide range of
> machines.
>
> I can happily build DLLs on any of our many machines that run on all
the
> others, and I use __cdecl, with M$ VC++, Watcom C and Fortran 10.0,
10.6,
> 11.0 and several others. There are problems with calling some of the
> functions in Sqpe.dll, but only because the conventions for returning
> values are not defined in the DLL specs. (I have found and published
> workarounds: see the on-line programming complements to Venables and
> Ripley
> on my web site.)
>
> I do wonder why your experience is so different from mine.
>
> --
> Brian D. Ripley, ripley@stats.ox.ac.uk
> Professor of Applied Statistics, http://www.stats.ox.ac.uk/~ripley/
> University of Oxford, Tel: +44 1865 272861 (self)
> 1 South Parks Road, +44 1865 272860 (secr)
> Oxford OX1 3TG, UK Fax: +44 1865 272595
[Clark Sims]
I did not mention the dyn.load or the compile.bat because they
work
without any problems. However they are limiting in the sense that you
can
only use them with C function as opposed to C++ functions. The are also
limiting in that you can not use the source level debugger. It is very
difficult to debug a large C/C++ function without a source level
debugger.
In regards to __cdecl, I have had some success in getting the
__cdecl to work, however it is not reliable. Steve Willer at tech
support
from Mathsoft originally pointed my in the direction of using only
__stdcall. Apparently Mathsoft is aware of the problem but they have not
adjusted their documentation accordingly.
I enjoyed your book on Modern Applied Statistics.
Best Regards,
Clark Sims
-----------------------------------------------------------------------
This message was distributed by s-news@wubios.wustl.edu. To unsubscribe
send e-mail to s-news-request@wubios.wustl.edu with the BODY of the
message: unsubscribe s-news
-----------------------------------------------------------------------
This message was distributed by s-news@wubios.wustl.edu. To unsubscribe
send e-mail to s-news-request@wubios.wustl.edu with the BODY of the
message: unsubscribe s-news
|