Questions about Ada Core Technologies

Questions about Ada Core Technologies

[Ludovic Brenta]

Hi

I got some flak[1] about things I wrote in my draft Debian Policy for
Ada.  While I prepare Draft 2, I would like to get some information in
order to get some facts straight.  Here is a list of questions I have;
please answer them only if you have first-hand knowledge; I am trying
to dispel rumours and spread facts :)

- Will ACT make more "p" releases of GNAT in the future?  They told me
  privately they would, but has anyone else heard about a public
  statement from ACT?  Is it too much to ask for a release date or
  time frame?

- Have ACT really switched their day-to-day development to the FSF?
  The changelogs suggest so, in which case I can suppose they merge
  selected changes to GNAT Pro in their private repository?

- Will the next GNAT Pro be based on FSF's GCC, or on ACT's private
  repository?

- Does ACT recommend anyone switch to GCC instead of GNAT 3.15p?  If
  so, which version of GCC?

- Does ACT request that customers not distribute copies of GNAT Pro?

- Since GNAT Pro, as a derivative work from GCC, is necessarily
  distributed under the GPL, is the above request not an infringement
  of the GPL?

- Just out of historical curiosity, could someone send me a timeline
  of GNAT Pro releases?  Especially the ones after 3.15p went out: I
  heard about 3.16a, 3.16a1, 3.17w, 5.01?, 5.02? etc.  and I would
  like to know what version of GCC they used as a backend.

[1] I mean I got *friendly* flak, kind of like error messages from an
Ada compiler :)

-- 
Ludovic Brenta.

Re: Questions about Ada Core Technologies

[Stephen Leake]

Ludovic Brenta <ludovic.brenta@insalien.org> writes:

I'm a paying customer of ACT, so I know the answer to some of these
questions.

> - Will ACT make more "p" releases of GNAT in the future?  They told me
>   privately they would, but has anyone else heard about a public
>   statement from ACT?  

The last I heard from them on this topic (about 6 months ago) was that
they had not decided.

>   Is it too much to ask for a release date or time frame?

Yes, it is too much to ask :).

> - Have ACT really switched their day-to-day development to the FSF?

I don't know. I also don't see how this is relevant to Debian.

> The changelogs suggest so, in which case I can suppose they merge
> selected changes to GNAT Pro in their private repository?

I assume they are maintaining at least a separate branch, if not a
separate repository, but I really don't know.

> - Will the next GNAT Pro be based on FSF's GCC, or on ACT's private
> repository?

5.02a is current (released in March 2004). It is based on FSF gcc
3.2.3 (that's what gcc --version says).

As usual, ACT has made no firm statments about future plans.

> - Does ACT recommend anyone switch to GCC instead of GNAT 3.15p? If
> so, which version of GCC?

I believe they would say "test it with your application; use whichever
is best for you".

> - Does ACT request that customers not distribute copies of GNAT Pro?

No. To be specific, there is nothing in the support contract that says
this. They do point out that the non-public releases are non-public
for a reason; they are more likely to contain bugs, and therefore
should only be used with a support contract. Customers tend to agree
with that position.

Again, I don't see how this is relevant to Debian.

> - Since GNAT Pro, as a derivative work from GCC, is necessarily
> distributed under the GPL, is the above request not an infringement
> of the GPL?

In short, the GPL says "If you give someone a binary, you also have to
give them the source". 

The GPL does _not_ say "if you give one person a binary, you also have
to give everyone else the same binary".

So even if ACT was requesting that customers do not distribute
non-public releases, it would not be violating the GPL

> - Just out of historical curiosity, could someone send me a timeline
> of GNAT Pro releases? Especially the ones after 3.15p went out: I
> heard about 3.16a, 

Feb 2003, gcc 2.8.1

> 3.16a1, 

June 2003, gcc 2.8.1

> 3.17w, 

The "w" means it was a "wavefront"; that's a bug fix for a particular
bug; not an official release.

> 5.01?, 

5.01a July 2003, gcc 3.2.3

> 5.02? 

5.02a March 2004, gcc 3.2.3


> etc. and I would like to know what version of GCC they used as a
> backend.
> 
> [1] I mean I got *friendly* flak, kind of like error messages from an
> Ada compiler :)

The latest version of your Debian policy looks good to me, except for
the part about ACT requesting non-distribution of non-public versions.

-- 
-- Stephe

Re: Questions about Ada Core Technologies

[Steve]

First:
  The best way to get these questions answered is ask ACT.

Second:
  I monitor the gcc mailing list (gcc.gnu.org/ml/gcc), and believe I know
the answers to some, but not all of these questions from monitoring that
list.


"Ludovic Brenta" <ludovic.brenta@insalien.org> wrote in message
news:87oeq4vkod.fsf@insalien.org...
> Hi
>
> I got some flak[1] about things I wrote in my draft Debian Policy for
> Ada.  While I prepare Draft 2, I would like to get some information in
> order to get some facts straight.  Here is a list of questions I have;
> please answer them only if you have first-hand knowledge; I am trying
> to dispel rumours and spread facts :)
>
> - Will ACT make more "p" releases of GNAT in the future?  They told me
>   privately they would, but has anyone else heard about a public
>   statement from ACT?  Is it too much to ask for a release date or
>   time frame?

The last I heard is that they have no plans to stop making tese releases.

> - Have ACT really switched their day-to-day development to the FSF?
>   The changelogs suggest so, in which case I can suppose they merge
>   selected changes to GNAT Pro in their private repository?

Of course ACT will be maintaining their own site.  They have customers that
depend on them and they cannot tolerate instability.

A few months ago there was an announcement on the gcc mailing list that ACT
would be updating the FSF repositoriy on a regular basis.  If I recall
correctly, Arnauld Charlet of ACT was made responsible for keeping the FSF
repository up to date.

> - Will the next GNAT Pro be based on FSF's GCC, or on ACT's private
>   repository?

ACT's private repository.  Over time the difference between the two will
continue to decrease.

> - Does ACT recommend anyone switch to GCC instead of GNAT 3.15p?  If
>   so, which version of GCC?
>

ACT doesn't recommend anyone use a 'p' release for serious work.  The public
releases are primarily released for academic use.  Of course the 'p'
releases are of good quality and could be used for serious work, but ACT
does not stand behind them.  For that you need to by GNAT Pro.

> - Does ACT request that customers not distribute copies of GNAT Pro?
>
> - Since GNAT Pro, as a derivative work from GCC, is necessarily
>   distributed under the GPL, is the above request not an infringement
>   of the GPL?
>
> - Just out of historical curiosity, could someone send me a timeline
>   of GNAT Pro releases?  Especially the ones after 3.15p went out: I
>   heard about 3.16a, 3.16a1, 3.17w, 5.01?, 5.02? etc.  and I would
>   like to know what version of GCC they used as a backend.
>

This has come up many times.  No.  You cannot get a time line for a public
release.
You can look at the GCC scheduled time line for releases, but you will find
that they too favor stability over deadlines.

I hope this helps,
Steve
(The Duck)

> [1] I mean I got *friendly* flak, kind of like error messages from an
> Ada compiler :)
>
> -- 
> Ludovic Brenta.

Re: Questions about Ada Core Technologies

[Florian Weimer]

Ludovic Brenta <ludovic.brenta@insalien.org> writes:

> - Have ACT really switched their day-to-day development to the FSF?
>   The changelogs suggest so, in which case I can suppose they merge
>   selected changes to GNAT Pro in their private repository?

The main development happens on ACT's infrastructure (after all, they
have a far more extensive test suite (which is partly covered by NDAs
and cannot be published), an automated regression tester based on that
test suite, a separate bug tracking system, their own developer
communication channels &c).

Arnaud Charlet regularly (daily?) merges ACT's changes into the FSF
tree.  But this doesn't mean that this tree is the sole reference for
development.

-- 
Current mail filters: many dial-up/DSL/cable modem hosts, and the
following domains: postino.it, tiscali.co.uk, tiscali.cz, tiscali.it,
voila.fr.

Re: Questions about Ada Core Technologies

[Jacob Sparre Andersen]

Stephen Leake wrote:

> So even if ACT was requesting that customers do not distribute
> non-public releases, it would not be violating the GPL

Ignoring the difference between "requesting" and "requiring", it
_would_ be a violation of the GNU GPL, since the GNU GPL allows people
to redistribute binaries (as long as they also redistribute the
corresponding source code).

But since FSF and ACT seem to be on friendly terms, I doubt that ACT
is doing anything that isn't completely by the book, when it comes to
GNU GPL.

Jacob
-- 
�And what about homo sapiens?
 Yes, we think that would be a very good idea ...� -- not Gandhi

Re: Questions about Ada Core Technologies

[Stephen Leake]

Jacob Sparre Andersen <sparre@nbi.dk> writes:

> Stephen Leake wrote:
> 
> > So even if ACT was requesting that customers do not distribute
> > non-public releases, it would not be violating the GPL
> 
> Ignoring the difference between "requesting" and "requiring", it
> _would_ be a violation of the GNU GPL, since the GNU GPL allows people
> to redistribute binaries (as long as they also redistribute the
> corresponding source code).

Hmm. You didn't quote my full post, so you've lost the definition of
"it" that we  are discussing here. Here is the original statement:

> - Does ACT request that customers not distribute copies of GNAT Pro?

Hmm. Your question caused me to go re-read the GPL. Here's the
relevant paragraphs:

 1. You may copy and distribute verbatim copies of the Program's
    source code as you receive it, in any medium, provided that you
    conspicuously and appropriately publish on each copy an
    appropriate copyright notice and disclaimer of warranty; keep
    intact all the notices that refer to this License and to the
    absence of any warranty; and give any other recipients of the
    Program a copy of this License along with the Program.

 3. You may copy and distribute the Program (or a work based on it,
    under Section 2) in object code or executable form under the terms
    of Sections 1 and 2 above provided ...

That does indeed say that if you have a copy of the binary or source,
you may distribute it. Note that it does _not_ say you _must_
distribute it, which some people seem to think, sometimes.

So I stand corrected; ACT may not _require_ that customers not give
copies of GNAT to anyone. However, they may _request_ it, for the "not
ready to distribute" reason I gave.

> But since FSF and ACT seem to be on friendly terms, I doubt that ACT
> is doing anything that isn't completely by the book, when it comes
> to GNU GPL.

Right. As I said, there is nothing in the customer contract with ACT
about not distributing GNAT. But they do informally request it.

-- 
-- Stephe

Re: Questions about Ada Core Technologies

[Robert I. Eachus]

Stephen Leake wrote:

> Right. As I said, there is nothing in the customer contract with ACT
> about not distributing GNAT. But they do informally request it.

Actually, what ACT requests/recommends is not distributing wavefront 
versions.  Wavefront versions are usually a response to a particular 
problem encountered by a (supported) user.  As such they are less 
thoroughly tested than other versions of GNAT.

What about the versions ending in a, not w?  They are betwixt and 
between.  In general if an a version gets to where it has a short 
buglist and no outstanding major problems, it gets converted to a p 
(public) version pretty quickly.  What has been going on recently, as I 
understand it, is that ACT has versions which perform well on some 
platforms but fail on others.  When you have customers with a support 
contract, you can insure that they have a variant that runs on the 
hardware (and OS) that customer uses.

Once the 5.xx versions become stable, ACT will stop recommending that 
the public use 3.15p.  Notice that the 5.xx versions are completely 
available to the public, just not recommended unless ACT has insured 
that they work in your environment.  (And of course they are under no 
obligation to test those versions on non-supported hardware and OS 
configurations.)

As for me, I have been staying with 3.15 because I want the software I 
write to be as widely distributable as possible.  (I just went through a 
painful process with a failing disk.  If it had failed completely I 
would have just gone to my last complete backup.  But when it started 
failing I was able to do one last incremental backup--but it took days. 
  Then I had to build a new disk, and apply all the increments.  Since 
that disk held the partition I use for Ada development, I didn't want to 
change any libraries in the middle of all this.)


-- 

                                           Robert I. Eachus

"The terrorist enemy holds no territory, defends no population, is 
unconstrained by rules of warfare, and respects no law of morality. Such 
an enemy cannot be deterred, contained, appeased or negotiated with. It 
can only be destroyed--and that, ladies and gentlemen, is the business 
at hand."  -- Dick Cheney

Re: Questions about Ada Core Technologies

[Peter Hermann]

Robert I. Eachus <rieachus@comcast.net> wrote:
> As for me, I have been staying with 3.15 because I want the software I 
> write to be as widely distributable as possible.  (I just went through a 

so do I:
because I want to be at eye level compatibility with my students.

I report a sad reaction from ACT:
I wanted to pay for a gnat3.15-compiler on my pet machine DECalpha
where I am frozen at gnat3.12p
without any need of support
but ACT rejected the deal.
Maybe DECalphaTrue64 is particularly difficult for the GNAT-compiler?

-- 
--Peter Hermann(49)0711-685-3611 fax3758 ica2ph@csv.ica.uni-stuttgart.de
--Pfaffenwaldring 27 Raum 114, D-70569 Stuttgart Uni Computeranwendungen
--http://www.csv.ica.uni-stuttgart.de/homes/ph/
--Team Ada: "C'mon people let the world begin" (Paul McCartney)

Re: Questions about Ada Core Technologies

[Robert I. Eachus]

Peter Hermann wrote:

> I report a sad reaction from ACT:
> I wanted to pay for a gnat3.15-compiler on my pet machine DECalpha
> where I am frozen at gnat3.12p
> without any need of support
> but ACT rejected the deal.
> Maybe DECalphaTrue64 is particularly difficult for the GNAT-compiler?

No, they probably don't want to get drawn in to supporting older 
versions of the compiler.  If they fixed a bug in 3.15 for you, would 
they also fix it in 3.15p?

However, if you really want a 3.15p compiler on you DEC Alpha, get the 
3.13p, 3.14p, and 3.15p sources and build it.  It may not be necessary 
to go through the intermediate compiler versions, but I think the time 
you would save would not be worth it.  You also may want to run the 
ACATS test suite against each version to look for regressions.

The whole process might take a week, but if you set up a different 
account for it, your involvement could be just a few hours worth.  For 
each iteration, you are going to be compiling the compiler and test 
suite twice.  Having a script set up to do that should allow you to do 
one iteration a day, with your involvement being checking on progress, 
and seeing if there are any regressions to worry about.  (You could also 
have to worry about different GCC versions.  Same procedure, different 
compilers.)

-- 

                                           Robert I. Eachus

"The terrorist enemy holds no territory, defends no population, is 
unconstrained by rules of warfare, and respects no law of morality. Such 
an enemy cannot be deterred, contained, appeased or negotiated with. It 
can only be destroyed--and that, ladies and gentlemen, is the business 
at hand."  -- Dick Cheney