From Keith.Bierman@eng.sun.com Thu Mar 2 02:34:56 1995 Received: from Sun.COM (koriel.Sun.COM) by dkuug.dk with SMTP id AA13127 (5.65c8/IDA-1.4.4j for ); Thu, 2 Mar 1995 23:16:04 +0100 Received: from Eng.Sun.COM (engmail2.Eng.Sun.COM) by Sun.COM (sun-barr.Sun.COM) id AA26095; Thu, 2 Mar 95 14:14:55 PST Received: from chiba.Eng.Sun.COM by Eng.Sun.COM (5.x/SMI-5.3) id AA25223; Thu, 2 Mar 1995 10:35:46 -0800 Received: from localhost by chiba.Eng.Sun.COM (4.1/SMI-4.1) id AA06968; Thu, 2 Mar 95 10:34:57 PST Message-Id: <9503021834.AA06968@chiba.Eng.Sun.COM> To: Miles Ellis Cc: SC22WG5@dkuug.dk Subject: personal submission for the april meeting Date: Thu, 02 Mar 95 10:34:56 PST From: Keith.Bierman@eng.sun.com X-Charset: ASCII X-Char-Esc: 29 I do not recall a virtual meeting of the subgroup on inter-operability having occurred. So the following remarks should be taken as my personal position only. Currently virtually every operating environment of any commercial consequence has services which are defined (or are readily definable) in terms of a C binding. While Objective-C and C++ may take over in the future, they are extensions to C, not replacements. Therefore ensuring that a Fortran program can call an arbitrary C "routine" would do much to facilitate the usage of Fortran. Calling arbitrary C routines requires some extensions to Fortran. We need 1) conformant datatypes bitfields unsigned c strings specified mappings (perhaps a standard module which provides the platform specific binding, but with standardized spellings, such as kinds C_FLOAT, C_DOUBLE, etc.) 2) calling conventions specifically STDARG mapping name mapping (the Fortran name may not be the C name, e.g. _foo->foo) character*(*) etc. If we had these features in the Standard, other groups (end users, vendors, industry consortia) could easily provide bindings to *all* system services (posix, posix threads, windowing systems, network services, etc). It is impossible for us (Fortran standards workers) to replicate the work of the larger wider community. Nor is it desirable to seek out lowest common denominators (e.g. one could create a "portable Fortran" subset of popular windowing systems such as WSWIN32, X11R5, X11R6, and Display Postscript. But the result would make users unhappy on *any* of those platforms. But with appropriate additions to our Standard, it would be possible for each of those technology suppliers (MS, X-consortium, Adobe) to provide a Fortran binding. If they chose not to, users could do the work themselves). I propose that WG5 establish a subgroup whose charter is narrow and timely. The group should prepare a report which details a complete facility for binding to the ISO C language. After we have such a report, we could ballot it as a technical report *and* incorporate it into the F2K base document. Thus it would have sufficient standing to be worthwhile for implementors to implement and users to eventually rely upon. There will, no doubt, be those who would prefer that we align ourselves with more grand projects such as ORB or CLID/CLIP or any of a dozen other efforts. I do not believe that such work will prove timely, or in some cases be efficient enough, or in many cases ever actually be widely implemented. One can argue that this does not support certain other ISO efforts; but it is vital to our user community. I feel the latter is the community we ought to serve in this matter.