Xref:	news-dnh.mv.net comp.os.msdos.djgpp:1763
Path:	news-dnh.mv.net!mv!news.sprintlink.net!in1.uu.net!gwu.edu!gwu.edu!not-for-mail
From:	mfeldman AT seas DOT gwu DOT edu (Michael Feldman)
Newsgroups:	comp.os.msdos.djgpp
Subject:	GNAT Ada 95 for GCC; support for same
Date:	25 Aug 1995 22:53:17 -0400
Organization:	George Washington University
Lines:	150
Nntp-Posting-Host:	128.164.9.3
To:	djgpp AT sun DOT soe DOT clarkson DOT edu
Dj-Gateway:	from newsgroup comp.os.msdos.djgpp

For some reason I did not see the attached message go by; someone
sent it to me by e-mail, suggesting that I might like to respond.

I do so after the extracts. This will be just a bit long, so bear with me.

> >In article <40n13f$qn7 AT phcoms4 DOT seri DOT philips DOT nl>, W DOT Derks AT nl DOT cis DOT philips DOT com
> >(Wiljan Derks) wrote:

> > In article <40ke9k$mdb AT freenet DOT vancouver DOT bc DOT ca>, 
> > rdc AT freenet DOT vancouver DOT bc DOT ca says...
> > >
> > >Just got GNAT to learn Ada on - and some tutorials.
> > >Curious if any other are using the Ada frontend for Djgpp's GCC.
> > >
{some stuff deleted]

> > For everybody using C or C++ right now I can just say: switch to GNAT
> > as soon as possible.
> Hmm.. contentious statement.  Ada is a f***ing HUGE language and takes
> quite a while to learn, and GNAT doesn't implement all features (esp.
> tasking) on Dos yet.  Also, I can't find a published Ada95 reference
> manual, which I consider a necessary "at hand" thing for looking up those
> constructs.  still, an Ada83 one will do in the meantime.

> > 
> > Notice that GNAT is not just a front end but a real compiler. It is linked 
> > agains the gcc backend however.

> indicating that it's not a compiler.  One of the papers published by the
> team developing GNAT tells you that - the name stands for GNU NYU Ada
> Translator.  it just turns Ada source code into the required form for the
> GCC back end, so it really is just a front end.

> Cheers

> Richard

We at GW have been responsible for the GNAT port to DOS for the last few
months, so we are somewhat familiar with it and with Ada 95.

First, your point about GNAT not being a compiler but "only" a front
end. Naturallly it is. _All_ the GCC compilers are just front ends.
C, C++, Objective C, etc. That is, in fact, the whole idea behind GCC -
multiple language front ends, multiple machine descriptions in the
back end. 

So GNAT is, in fact, just another GCC compiler (front end,
whatever), just like the others. The only difference is that it is not
written in C, but in itself, mostly Ada 95 with a bit of C around the 
edges. This makes no difference for users, only for those who want to
build it from sources - you need a GNAT to compile GNAT, so the first 
version of a port to a new platform is done by cross-compiling.

   * * * * * *

Second - the size of Ada in general - sure, Ada is large, but no larger
than other "big" languages like C++ or Modula-3. Since there is no agreed
metric for "size" of languages, arguing this point just gets into the
sort of religious war I'd like to avoid (on this group, anyway:-)).

It is also the case that, as with any large language, there's no need to
swallow it all in one gulp. For a list of good texts and other tutorial
stuff, see the web page at http://lglwww.epfl.ch/Ada/ or e-mail me.

   * * * * * *

Third - Ada 95 reference manual. If you really want hard copy, you can
order it through the Ada Information Clearinghouse. I can give more
details on this by e-mail.

OTOH, most of us are finding that it's just as well to use the Ada 95 LRM
in electronic form. It is available in various forms including ASCII,
PostScript (read with GhostScript or print it yourself!), HTML, and
Macintosh HyperCard. The best site for this is wuarchive.wustl.edu,
in the directory languages/ada. Also check the Ada web site,
http://lglwww.epfl.ch/Ada/.

There is also a hypertext Ada 95 LRM (not HTML, a homebrewed simple 
hypertext system) included in the AdaCAPS IDE for GNAT/DOS, released in July.
I will post an announcement of this, and how to get an easy-to-install
distribution of GNAT and other stuff, in a separate note. In brief,
try ftp.gwu.edu, pub/ada/ez2load for all the stuff.

The Ada 95 standard, like the Ada 83 one, is just about the only language
standard available free on the net and freely copiable.

   * * * * * *

Fourth - tasking in DOS. It is true that the runtime support for tasking
under DOS is still under development. Charles Sandmann is involved in
this effort, which is trying to put a good threads package into djgpp,
for use with GNAT but also with the other GCC languages. A student of mine
and another at Florida State (where the GNAT tasking system is produced)
are also involved. We hope this will bear fruit before long. Meanwhile,
if any of you are running OS/2, the GNAT port for OS/2 does tasking 
just fine. Pick it up from cs.nyu.edu, pub/gnat.

   * * * * * *

Finally, someone asked in another note whether support is available for
a fee for GNAT. The original NYU group that developed GNAT has moved
into a "commercial spinoff" called Ada Core Technologies or ACT.
This company is built on the cygnus model, to support GNAT specifically.
I am not connected with this company; I just know the people.:-)
Write to them at report AT gnat DOT com to inquire about support; I'm sure
they will be happy to discuss your needs.

For your interest: ACT has a fair amount of funding from Silicon Graphics,
which has decided that GNAT will be its Ada 95 compiler. SGI is helping
ACT get GNAT ready for formal validation (against the government-sponsored
test suite), which should happen in the next few months. SGI has also
released some really nice (proprietary, not free) CASE tools and other
stuff that fits around GNAT on the SGI, and has shown some really wifty
virtual-reality-type demos using GNAT and C++ talking to each other. I
mention this only to indicate that GNAT has a growing amount of industry
support from major companies, and it isn't even quite done.:-)

Also, as we know about GCC tools in general, portability of GNAT is
quite high, so a validation on one platform means validation on others
will not be difficult to accomplish. Because even the unvalidated ports
use the identical front end, we can have confidence that GNAT will be
a good representation of the intent of the standard, which is what 
validation is all about.

One last point: ACT has reminded everyone that GNAT is and will
remain free software under the GPL/LGPL, and will remain freely
available with or without support.

   * * * * * *

Sorry for the length of this; I thought I'd put it all in one place.
Feel free to get in touch by e-mail if you want more details. We are
planning a listproc-based mailing list for general (non-ACT) GNAT
users, and will announce its formation here in about 2 weeks. My plan
is to let it run like the old djgpp list, then if traffic gets heavy,
put out a CFV for a newsgroup.

Cheers, all -

Mike Feldman
------------------------------------------------------------------------
Michael B. Feldman -  chair, SIGAda Education Working Group
Professor, Dept. of Electrical Engineering and Computer Science
The George Washington University -  Washington, DC 20052 USA
202-994-5919 (voice) - 202-994-0227 (fax) - mfeldman AT seas DOT gwu DOT edu (Internet)
------------------------------------------------------------------------
                         Fight FUD with Fact
------------------------------------------------------------------------
Ada on the WWW: http://lglwww.epfl.ch/Ada/ or http://info.acm.org/sigada/
------------------------------------------------------------------------

- Raw text -