comp.lang.ada
 help / color / mirror / Atom feed
* GNAT 3.3 vs 3.15
@ 2003-04-01 16:51 David Holm
  2003-04-01 20:19 ` Stephen Leake
  0 siblings, 1 reply; 6+ messages in thread
From: David Holm @ 2003-04-01 16:51 UTC (permalink / raw)


This has surely been answered before but I didn't find a satisfactory answer to it on
groups.google.com.

What are the problems of using GNAT 3.3 over 3.15? What is new since 3.15?
I've been using 3.3 for a while and run it through my test suites (some simple floating-point
stuff) and so far have had nog problems.

//David Holm



^ permalink raw reply	[flat|nested] 6+ messages in thread

* Re: GNAT 3.3 vs 3.15
  2003-04-01 16:51 GNAT 3.3 vs 3.15 David Holm
@ 2003-04-01 20:19 ` Stephen Leake
  2003-04-01 21:18   ` David Holm
                     ` (2 more replies)
  0 siblings, 3 replies; 6+ messages in thread
From: Stephen Leake @ 2003-04-01 20:19 UTC (permalink / raw)


David Holm <david@realityrift.com> writes:

> What are the problems of using GNAT 3.3 over 3.15? 

I'm not clear what you mean by "GNAT 3.3"; there is no such thing
according to ACT. I guess you mean "The Ada compiler that comes with
GCC 3.3". What does "gnatmake -v" say about the version?

> What is new since 3.15? 

Bugs. The main difference is the gcc 3.x backend; it has a lot of
changes since the gcc 2.8 that gnat 3.15 is based on. Some of those
changes break GNAT. To be fair, some of them improve GNAT :). ACT is
working on fixing things, but it's a slow process.

For example, the ACT candidate release of GNAT based on gcc 3.x
(called GNAT 5.00a) is now available for customer testing; it has a
bug that stopped me cold. They are fixing it, but I won't be using
GNAT 5.00 (or GCC 3.3) until there is another release from ACT.

> I've been using 3.3 for a while and run it through my test suites
> (some simple floating-point stuff) and so far have had nog problems.

Then you are ok; that's what tests are for, after all!

You should improve your test suite to truly test your application;
that may make you feel more comfortable.

But keep GNAT 3.15 handy; if you encounter a weird problem, try it on
3.15 before you blame your code :).

-- 
-- Stephe



^ permalink raw reply	[flat|nested] 6+ messages in thread

* Re: GNAT 3.3 vs 3.15
  2003-04-01 20:19 ` Stephen Leake
@ 2003-04-01 21:18   ` David Holm
  2003-04-02  3:39     ` Steve
  2003-04-02 12:01   ` Preben Randhol
  2003-04-04 10:55   ` Arne Knutsen
  2 siblings, 1 reply; 6+ messages in thread
From: David Holm @ 2003-04-01 21:18 UTC (permalink / raw)


On 01 Apr 2003 15:19:33 -0500
Stephen Leake <Stephen.A.Leake@nasa.gov> wrote:

> David Holm <david@realityrift.com> writes:
> 
> > What are the problems of using GNAT 3.3 over 3.15? 
> 
> I'm not clear what you mean by "GNAT 3.3"; there is no such thing
> according to ACT. I guess you mean "The Ada compiler that comes with
> GCC 3.3". What does "gnatmake -v" say about the version?

Um, yes, I was under the impression that ACT wouldn't release more gnat versions but instead go
through gcc.
gnatmake -v says:
GNATMAKE 3.3 20030329 (prerelease) Copyright 1995-2002 Free Software Foundation, Inc.

> > What is new since 3.15? 
> 
> Bugs. The main difference is the gcc 3.x backend; it has a lot of
> changes since the gcc 2.8 that gnat 3.15 is based on. Some of those
> changes break GNAT. To be fair, some of them improve GNAT :). ACT is
> working on fixing things, but it's a slow process.
> 
> For example, the ACT candidate release of GNAT based on gcc 3.x
> (called GNAT 5.00a) is now available for customer testing; it has a
> bug that stopped me cold. They are fixing it, but I won't be using
> GNAT 5.00 (or GCC 3.3) until there is another release from ACT.
> 
> > I've been using 3.3 for a while and run it through my test suites
> > (some simple floating-point stuff) and so far have had nog problems.
> 
> Then you are ok; that's what tests are for, after all!
> 
> You should improve your test suite to truly test your application;
> that may make you feel more comfortable.
> 
> But keep GNAT 3.15 handy; if you encounter a weird problem, try it on
> 3.15 before you blame your code :).
> 
> -- 
> -- Stephe



^ permalink raw reply	[flat|nested] 6+ messages in thread

* Re: GNAT 3.3 vs 3.15
  2003-04-01 21:18   ` David Holm
@ 2003-04-02  3:39     ` Steve
  0 siblings, 0 replies; 6+ messages in thread
From: Steve @ 2003-04-02  3:39 UTC (permalink / raw)



"David Holm" <david@realityrift.com> wrote in message
news:20030401231809.4e51c34a.david@realityrift.com...
> On 01 Apr 2003 15:19:33 -0500
> Stephen Leake <Stephen.A.Leake@nasa.gov> wrote:
>
> > David Holm <david@realityrift.com> writes:
> >
> > > What are the problems of using GNAT 3.3 over 3.15?
> >
> > I'm not clear what you mean by "GNAT 3.3"; there is no such thing
> > according to ACT. I guess you mean "The Ada compiler that comes with
> > GCC 3.3". What does "gnatmake -v" say about the version?
>
> Um, yes, I was under the impression that ACT wouldn't release more gnat
versions but instead go
> through gcc.
> gnatmake -v says:
> GNATMAKE 3.3 20030329 (prerelease) Copyright 1995-2002 Free Software
Foundation, Inc.
>
[snip]

The last I heard ACT will continue to make pre-built public releases of GNAT
for a few platforms.  For other platforms you will have to do your own
build.

Version number for GNAT are independent of version number of GCC.

FYI: The last I heard:
GNAT Compilers version 3.x are based on GCC 2.8.1
GNAT Compilers version 4.x validated (don't know if they specify base GCC
for these)
GNAT Compilers version 5.x are based on GCC 3.x

A recent post on the GCC mailing list indicated that ACT will soon be set up
to keep the GCC CVS tree more in line with their internal repository.  This
is good news.

Steve
(The Duck)





^ permalink raw reply	[flat|nested] 6+ messages in thread

* Re: GNAT 3.3 vs 3.15
  2003-04-01 20:19 ` Stephen Leake
  2003-04-01 21:18   ` David Holm
@ 2003-04-02 12:01   ` Preben Randhol
  2003-04-04 10:55   ` Arne Knutsen
  2 siblings, 0 replies; 6+ messages in thread
From: Preben Randhol @ 2003-04-02 12:01 UTC (permalink / raw)


Stephen Leake wrote:
> David Holm <david@realityrift.com> writes:
> 
>> What are the problems of using GNAT 3.3 over 3.15? 
> 
> I'm not clear what you mean by "GNAT 3.3"; there is no such thing
> according to ACT. I guess you mean "The Ada compiler that comes with
> GCC 3.3". What does "gnatmake -v" say about the version?

gnat-3.3 is the gnat compiler for the gcc 3.3 backend. Maybe ACT will
call the supported version 5.0, but I very much doubt that the unsupported
gnat compiler won't be versioned 3.x according to the gcc version.

Preben



^ permalink raw reply	[flat|nested] 6+ messages in thread

* Re: GNAT 3.3 vs 3.15
  2003-04-01 20:19 ` Stephen Leake
  2003-04-01 21:18   ` David Holm
  2003-04-02 12:01   ` Preben Randhol
@ 2003-04-04 10:55   ` Arne Knutsen
  2 siblings, 0 replies; 6+ messages in thread
From: Arne Knutsen @ 2003-04-04 10:55 UTC (permalink / raw)



"Stephen Leake" <Stephen.A.Leake@nasa.gov> wrote in message
news:uhe9i7yy2.fsf@nasa.gov...
> David Holm <david@realityrift.com> writes:
>
> > What are the problems of using GNAT 3.3 over 3.15?
>
> I'm not clear what you mean by "GNAT 3.3"; there is no such thing
> according to ACT. I guess you mean "The Ada compiler that comes with
> GCC 3.3". What does "gnatmake -v" say about the version?
>
> > What is new since 3.15?
>
> Bugs. The main difference is the gcc 3.x backend; it has a lot of
> changes since the gcc 2.8 that gnat 3.15 is based on. Some of those
> changes break GNAT. To be fair, some of them improve GNAT :). ACT is
> working on fixing things, but it's a slow process.
>
> For example, the ACT candidate release of GNAT based on gcc 3.x
> (called GNAT 5.00a) is now available for customer testing; it has a
> bug that stopped me cold. They are fixing it, but I won't be using
> GNAT 5.00 (or GCC 3.3) until there is another release from ACT.

What kind of bug(s) is this. We're having trouble with a SW that runs fine
on 3.16, but fails (no crahes) on 5.0


>
> > I've been using 3.3 for a while and run it through my test suites
> > (some simple floating-point stuff) and so far have had nog problems.
>
> Then you are ok; that's what tests are for, after all!
>
> You should improve your test suite to truly test your application;
> that may make you feel more comfortable.
>
> But keep GNAT 3.15 handy; if you encounter a weird problem, try it on
> 3.15 before you blame your code :).
>
> --
> -- Stephe





^ permalink raw reply	[flat|nested] 6+ messages in thread

end of thread, other threads:[~2003-04-04 10:55 UTC | newest]

Thread overview: 6+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2003-04-01 16:51 GNAT 3.3 vs 3.15 David Holm
2003-04-01 20:19 ` Stephen Leake
2003-04-01 21:18   ` David Holm
2003-04-02  3:39     ` Steve
2003-04-02 12:01   ` Preben Randhol
2003-04-04 10:55   ` Arne Knutsen

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox