[Q] GNAT Success stories?

[Q] GNAT Success stories?

[Dave Wright]

Has anyone successfully used gnat on a medium (+60K lines) project?  We're 
considering using gnat on a Pentium running Solaris X86.  Any comments, 
suggestions, warnings are appreciated.

TIA,
Dave

---------------------------------------------------------------
Dave Wright			wdwright@ccgate.hac.com
Systems Engineer		Hughes Technical Systems Company

Re: [Q] GNAT Success stories?

[Samuel Tardieu]

>>>>> "Dave" == Dave Wright <wdwright@ccgate.hac.com> writes:

Dave> Has anyone successfully used gnat on a medium (+60K lines)
Dave> project?  We're considering using gnat on a Pentium running
Dave> Solaris X86.  Any comments, suggestions, warnings are
Dave> appreciated.

Well, GNAT's front-end which is written in Ada is much larger than 60K 
lines and gets compiled by GNAT itself :-) Moreover, Solaris x86 is a
target fully supported by ACT (see Robert Dewar's post a few days ago).

I've also seen medium/large military projects (+500K lines) compiled
with GNAT running successfully, so the size of your project is not
going to be a problem.

  Sam
-- 
Samuel Tardieu -- sam@ada.eu.org

Re: [Q] GNAT Success stories?

[Samuel Tardieu]

PS> I have an unsuccess story. I have developed an Ada-library for
PS> education purposes under Meridian with nested generics,
PS> heterogenous lists, etc.  In trying to compile it under GNAT I
PS> received a lot of different kind of "internal errors", messages
PS> from assembler and similar phenomena.  My next try led to Aonix'
PS> Object Ada and I succeeded prompt.

You don't even say what version of GNAT or ObjectAda you tried!

Anyway, seeing how you configured your newsreader (your name and
address don't even appear in the "From" line), one can safely guess
that you misconfigured GNAT or even forgot to install the assembler :)

  Sam
-- 
Samuel Tardieu -- sam@ada.eu.org

Re: [Q] GNAT Success stories?

[Robert A Duff]

In article <qw6iv1dhiv5.fsf@esmeralda.enst.fr>,
Samuel Tardieu  <sam@ada.eu.org> wrote:
>PS> I have an unsuccess story. I have developed an Ada-library for
>PS> education purposes under Meridian with nested generics,
>PS> heterogenous lists, etc.  In trying to compile it under GNAT I
>PS> received a lot of different kind of "internal errors", messages
>PS> from assembler and similar phenomena.  My next try led to Aonix'
>PS> Object Ada and I succeeded prompt.
>
>You don't even say what version of GNAT or ObjectAda you tried!

Both claim to be full Ada 95 compilers, in whatever version.
Why shouldn't he expect some Ada 83 code to work properly?

>Anyway, seeing how you configured your newsreader (your name and
>address don't even appear in the "From" line), one can safely guess
>that you misconfigured GNAT or even forgot to install the assembler :)

Hmm.  Seems like "blame the victim" to me.  Here's a guy who claims some
software is buggy.  And you don't believe him, because he doesn't know
how to configure his newsreader.  Well, I don't know how to configure my
newsreader, either (I just hope my internet provider does), but I feel
quite competent to point out bugs in GNAT, when they exist.

- Bob

Re: [Q] GNAT Success stories?

[Dan Lehman]

Bob Duff (obviously an Ada duffer) wrote (and I am NOT making this up!)

> Well, I don't know how to configure my newsreader, either ...

WHOA!
Here, in light of recent revelations of incompetence at the FBI lab(s),
which now send hope to many convicted by their *evidence*, Bob confesses
to newsreader naivete'!
Wow, imagine all the Approved Commentaries on which Bob voted that might
now be called back by WG-9 for reconsideration.  And didn't Bob work on
the Ada 95 standard, in the MRT?  --better re-review that as well, might
turn out to be only Ada 94, or less (--been wondering about Section 13 :-).
As it's well after April 1st, I am shocked.

---Dan
------- *

Re: [Q] GNAT Success stories?

[Robert Dewar]

Dave Wright says

<<Has anyone successfully used gnat on a medium (+60K lines) project?>>

Most of our customers are using GNAT on much larger projects, up to
millions of lines, but there are certainly success stories on smaller
projects as well. We will be publishing a series of GNAT success stories
in the near future.

Meanwhile you might check out the information on the Airfields project,
I believe that there is information on this at the Ada IC. This was the
first deployed Ada 95 project in the DoD, and it used GNAT (this was
deployed about two years ago if my memory is correct).

Robert Dewar
Ada Core Technologies

Re: [Q] GNAT Success stories?

[Robert Dewar]

Dave Wright asks

<<Has anyone successfully used gnat on a medium (+60K lines) project?  We're
considering using gnat on a Pentium running Solaris X86.  Any comments,
suggestions, warnings are appreciated.>>

I should also mention that Solaris x86 is now fully supported by 
Ada Core Technologies (and there are current commercial users of
this port).

Robert Dewar
Ada Core Technologies

Re: [Q] GNAT Success stories?

[Robert A Duff]

In article <5jm6qm$gem@news.ida.org>, Dan Lehman <lehman@ida.org> wrote:
>Bob Duff (obviously an Ada duffer) wrote (and I am NOT making this up!)
>
>> Well, I don't know how to configure my newsreader, either ...
>
>WHOA!
>Here, in light of recent revelations of incompetence at the FBI lab(s),
>which now send hope to many convicted by their *evidence*, Bob confesses
>to newsreader naivete'!
>Wow, imagine all the Approved Commentaries on which Bob voted that might
>now be called back by WG-9 for reconsideration.  And didn't Bob work on
>the Ada 95 standard, in the MRT?  --better re-review that as well, might
>turn out to be only Ada 94, or less (--been wondering about Section 13 :-).
>As it's well after April 1st, I am shocked.

:-)

Actually, I thought it *was* Ada 94.  The last dot of ink (or was it
toner?) dried on the official paper version in early December 1994.  ISO
took several months to make it official, though.  ;-)

- Bob

Re: [Q] GNAT Success stories?

[Robert Dewar]

Bob Duff said

<<Hmm.  Seems like "blame the victim" to me.  Here's a guy who claims some
software is buggy.  And you don't believe him, because he doesn't know
how to configure his newsreader.  Well, I don't know how to configure my
newsreader, either (I just hope my internet provider does), but I feel
quite competent to point out bugs in GNAT, when they exist.>>

What Sam was trying to say was that we often find people using extremely
old versions of GNAT (someone recently submitted a bug with a version
that was over two years old), and it is important that such people should
know that there have been some improvements since. It is also the case
that some people, not that many, but some, simply don't follow the]
installation instructions, and what appears to them to be bugs are in
fact just the result of incorrect installation. 

I would certainly hope that the Aonix compiler should be super easy to
install and use (I don't know myself because I don't run windows!) since
that was the whole point of the large investment the government made in
the educational compiler. That is not particularly a design point of 
GNAT -- certainly we cannot invest millions of dollars in the effort,
nevertheless we try to improve installation ease as we go along.

One thing to say again here is that chat@gnat.com is a good place to get
informal installation assistance. Subscribe by sending mail to
chat-request@gnat.com.

Re: [Q] GNAT Success stories?

[Robert Dewar]

>PS> I have an unsuccess story. I have developed an Ada-library for
>PS> education purposes under Meridian with nested generics,
>PS> heterogenous lists, etc.  In trying to compile it under GNAT I
>PS> received a lot of different kind of "internal errors", messages
>PS> from assembler and similar phenomena.

If you are getting messages from the assembler, then this almost
certainly means that your installation is incorrect. Most often this is
the result of not following the installation instructions, including those
in the readme file, very carefully. For example, if you have the wrong paths
set up and pick up the wrong assembler, this could certainly result in the
kind of thing you are seeing.

In general if you install GNAT and get a situation where "nothing works",
then it is almost certain that you have misinstalled. GNAT installation
needs to be done carefully following all the instructions.

You may find it useful to join the chat mailing list (send mail to
chat-request@gnat.com) for informal installation help.

If you are pretty sure your installation is correct, e.g. all the example
programs run fine, and you get an error of the kind you describe, and you
are using the most recent version of GNAT, you should send an error report
to report@gnat.com, following the suggested format in gnatinfo.txt.

P.S. I am not quite sure what heterogenous lists might mean, but if you
really mean lists with different types of data on them, then that can
only be achieved in Ada 83 using implementation dependent tricks, which
may or may not be portable. Of course this would not account for the kind
of behavior you are seeing!

Re: [Q] GNAT Success stories?

[Solymosi]

bobduff@world.std.com (Robert A Duff) writes: 

> Hmm.  Seems like "blame the victim" to me.  Here's a guy who claims some
> software is buggy.  And you don't believe him, because he doesn't know
> how to configure his newsreader.  Well, I don't know how to configure my
> newsreader, either (I just hope my internet provider does), but I feel
> quite competent to point out bugs in GNAT, when they exist.
> 
> - Bob
Thanks Bob to save my honor.

Re: [Q] GNAT Success stories?

[Solymosi]

Samuel Tardieu <sam@ada.eu.org> writes: 
> You don't even say what version of GNAT or ObjectAda you tried!
> 
> Anyway, seeing how you configured your newsreader (your name and
> address don't even appear in the "From" line), one can safely guess
> that you misconfigured GNAT or even forgot to install the assembler :)
That's right. I don't know how to configure programs like News or Gnat 
and I don't know their versions either. I let it do my labor ingenieur. 
I can only hope he does a good work.
My student have been working with GNAT (whatever version it is), 
so there must be an assembler with. 
Nevertheless the labor ingenieur must resign before messages like 
"internal error".

Re: [Q] GNAT Success stories?

[Nick Roberts]

Newsreader? What's a newsreader? Someone tell me how to tune my telly,
never how to configure b****y newsreaders! I remember the days ...

Nick :-)

Re: [Q] GNAT Success stories?

[Robert Dewar]

<<That's right. I don't know how to configure programs like News or Gnat
and I don't know their versions either. I let it do my labor ingenieur.
I can only hope he does a good work.
My student have been working with GNAT (whatever version it is),
so there must be an assembler with.
Nevertheless the labor ingenieur must resign before messages like
"internal error".>>

The requirement for installing GNAT is simple in theory: read and follow
all the installation instructions precisely and exactly.

In practice, these days, especially in the NT world, people inherit an
attitude from the Mac world that software should install itself easily,
and you should not have to worry about reading anything or being precise.
Certainly that's the way *I* install most Win95 software today.

GNAT is not a turnkey installation on Win 95, there are some details that
you have to pay careful attention to, and although the great majority of
people manage to install it without trouble (e.g. several of my freshmen
last semester installed it just fine without any help), some people do
run into trouble.

So far, it is has not been a major design goal to make the installation
of GNAT NT idiot-proof (and no, I am not saying you are an idiot if you
make a mistake -- I am just using a common term that very nicely captures
a desirable goal -- the idea behind that idiom is that if even an idiot
can install it, then you have truly reached the goal of simple installation).

Our primary NT goal is to support serious development, particularly in the
case of moving large scale Unix programs to NT. Consequently the focus has
been on solving functionality problems (like the nasty ld problem, which
generally hits ONLY large programs -- and which finally appears to be solved)

We have *not* partiularly conentrated on easy installation, since that is
not a priority for our customers. 

It sure would be nice to have NT/GNAT install as easily as possible, and
we will continue to work on this in a low priority mode. As you will
probably know, Mike Feldman proposed a couple of years ago to do some
development to make GNAT on NT and the Mac really super easy to use, but
no one was interested in funding this. Instead the funding went to
Academic Ada, which means that you have a choice, and given the amount
of that funding, I would assume (though don't know first hand, because
I personally run OS/2 :-) that Object Ada should be super easy to install,
since clearly it is a priority goal for that project.

It may well be that for simple student use, GNAT is not the ideal choice,
although it does have very good error messages for student use, and that
alone has caused it to be the teaching vehicle of choice at many places,
even in some military academies, where other compilers could have been
used at no charge.

P.S. regarding the "nasty ld problem", we will probably prepare an update
of the 3.09 release that fixes this and make this available in the near
future. Not many people run into this, but if you do it can be a 
significant annoyance ...

Robert Dewar
Ada Core Technologies

Re: [Q] GNAT Success stories?

[Robert Dewar]

Solymos said

<<My student have been working with GNAT (whatever version it is),
so there must be an assembler with.>>

Yes, of course there is, but it sounds as though you may have been picking
up the wrong assembler version. This could result if you did not have your
paths setup exactly right for example ...