(j3.2006) [Fwd: RE: Ada's Fortran interoperability AI]

[Van Snyder]

The Ada standard has a Fortran interoperability annex that contains
significant mistakes.  Those ought to be corrected.  It also has an
annex each for C and COBOL interoperability.  Neither the Fortran nor
COBOL interoperability annex refers to the C interoperability annex.

The 2012 Ada standard (reference manual), annotated reference manual,
and rationale, are online at

http://www.adaic.org/ada-resources/standards/ada12/

and perhaps elsewhere as well.

If one wishes to recommend a different approach from the one in
AI12-0058 to remedy the deficiencies in the 2012 Fortran
interoperability annex, it ought to be submitted to WG9.  The current
chair is Joyce Tokar <tokar at pyrrhusoft.com>, but I'm sure Randy Brukardt
<randy at rrsoftware.com> would accept it.

On Tue, 2017-02-21 at 20:17 -0500, Steve Lionel wrote:
> I think this document is taking the wrong approach. Instead it should
> build on the C interop features. Indeed, if Ada can "look like C",
> then the Fortran standard C interop features can be used.
> 
> I am especially troubled by the notion that an Ada and a Fortran
> implementation should "conspire" to support a shared interface for
> arrays, allocatables and such. We already have that in F2015 and
> several implementations of the C interop stuff exist. 
> 
> The stuff about kinds should instead refer to C types the way C
> interop does. 
> 
> Most important, any Fortran routine called by Ada must be
> interoperable (and any Ada procedure called by Fortran must be
> interoperable), so that one doesn't get into implementation specifics
> such as hidden arguments, calling mechanisms and name decoration.
> 
> I assume Ada has a C interoperability specification. That should be
> usable with Fortran along with some guidance for the programmer.
> 
> Steve
> 
> On 2/21/2017 8:08 PM, Van Snyder wrote:
> 
> > I mentioned at J3 meeting 212 that there is a proposal to amend Ada's
> > annex concerning Fortran interoperability to be compatible with Fortran
> > 2008, not to recommend things that were never part of any Fortran
> > standard, and not to recommend things that are processor dependent or
> > dependent upon processor command-line options used when compiling a
> > Fortran program.
> > 
> > This resulted in generation of what WG9 calles an "AI" or "Ada Issue."
> > 
> > Dan asked me to forward the AI.
> > 
> > I asked Randy Brukardt, a denizen of WG9, to send it to me.
> > 
> > It's attached.  From my perspective, it looks quite good.  Let me (or
> > Randy) know if you have comments on it.
> > 
> > Van
> > 
> > -------- Forwarded Message --------
> > From: Randy Brukardt <randy at rrsoftware.com>
> > Reply-to: <randy at rrsoftware.com>
> > To: Van.Snyder at jpl.nasa.gov
> > Subject: RE: Fortran interoperability AI
> > Date: Tue, 21 Feb 2017 16:07:14 -0600
> > 
> > You can always get these out of our database (see
> > http://www.ada-auth.org/AI12-SUMMARY.HTML, looking for Fortran will find it
> > more quickly than asking me). I've attached the most recent version.
> > 
> >                  Randy. 
> > 
> > > -----Original Message-----
> > > From: Van Snyder [mailto:Van.Snyder at jpl.nasa.gov] 
> > > Sent: Tuesday, February 21, 2017 3:55 PM
> > > To: randy at rrsoftware.com
> > > Subject: RE: Fortran interoperability AI
> > > 
> > > Randy:
> > > 
> > > Can you send me a copy of the AI?  The Fortran committee 
> > > asked me for it when I mentioned it in my liaison report last week.
> > > 
> > > Thanks,
> > > Van
> > 
> > 
> > _______________________________________________
> > J3 mailing list
> > J3 at mailman.j3-fortran.org
> > http://mailman.j3-fortran.org/mailman/listinfo/j3
> 
> 
> _______________________________________________
> J3 mailing list
> J3 at mailman.j3-fortran.org
> http://mailman.j3-fortran.org/mailman/listinfo/j3