s-news
[Top] [All Lists]

Re: [S] Re: S-News Digest V1 #519

To: Frank E Harrell Jr <fharrell@virginia.edu>
Subject: Re: [S] Re: S-News Digest V1 #519
From: Prof Brian D Ripley <ripley@stats.ox.ac.uk>
Date: Fri, 30 Jul 1999 17:23:21 +0100 (BST)
Cc: Michael Prager <Mike.Prager@noaa.gov>, s-news <s-news@wubios.wustl.edu>, ALFRED.H.BALCH@stl.Monsanto.com
In-reply-to: <14241.49122.460000.706353@gargle.gargle.HOWL>
Sender: owner-s-news@wubios.wustl.edu
On Fri, 30 Jul 1999, Frank E Harrell Jr wrote:

> 
> Michael - I stay away from DLLS because I need to call builtin-fortran
> modules that S+ provides for matrix operations.  I would appreciate
> some info about compilers that produce .obj files that work with S+ as
> well as Watcom.  Thanks -Frank

I am sure there are none that produce .obj files in Watcom format. Some
Fortran compilers can produce DLLs. One of the best available is the free
one, g77: see the V&R2 Programming Complements for how to use it. Another
perfectly viable option is to use f2c plus your favourite C compiler: this
does work with Visual C++.

You do need to be very careful with using DLLs that call inbuilt functions:
the calling conventions are not fully defined and Watcom and (e.g.) Visual
C++ do differ.

> Michael Prager writes:
>  > 
>  > At 7/30/99 at 02:30 AM, <ALFRED.H.BALCH@stl.Monsanto.com> wrote:
>  > 
>  > >Am I better off getting a copy of the Watcom compiler than a compiler that
>  > >will be continued-e.g. Microsoft ?
>  > 
>  > The Microsoft Fortran compiler was discontinued years ago, so that would 
> be 
>  > a poor choice.  During its existence, I saw frequent complaints of its 
>  > bugginess and lack of conformance to the Fortran standard.
>  > 
>  > The Watcom compiler was a good one, but it was FORTRAN (77), an older 
>  > version of the language, and thus lacked most of the interesting features 
>  > of modern Fortran (95), the most notable in the context of S being 
>  > intrinsic array operations.  I would not recommend that *anyone* purchase 
> a 
>  > FORTRAN 77 compiler anymore.

But those intrinsic array operations do not match the F77 data structures
used by S.

I have tried both using dyn.load and dll.load extensively, the latter with
about half a dozen different compilers. It is no accident that all the code
in my SWin collection used Watcom and dyn.load. If I could not buy
those (and I would to ease my job) I would use the egcs-1.1.2-mingw32
versions of gcc/g77. The latter work well (as their use for the Windows
version of the other S-like system shows.)

[Please don't ask me where to find all these bits: try an Internet search.]

-- 
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

-----------------------------------------------------------------------
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

<Prev in Thread] Current Thread [Next in Thread>