Answer of Request to AdaCore on licensing Status of GtkAda 2.4.0

Answer of Request to AdaCore on licensing Status of GtkAda 2.4.0

[M E Leypold]

Dear All,

To make it short: A member of the sales departement of AdaCore has
written a mail to me, which at least tries to stipulate that even the
older (pre 2005 versions) of GtkAda can only be used under the
GPL. Which can only mean: The linking exception has been revoked, that
happening at a unspecified time in the deep past and silently so.


The Details:
============

Distribution and Licensing History of GtkAda
--------------------------------------------

The distribution history of GtkAda as far as I can reconstruct it,
has been the following:

 - GtkAda 1.x and GtkAda 2.x up to and including version 2.4.0 have
   been distributed from libre.act-europe.fr with a GMGPL license
   notice at the web site and a not quite so clear license notice in
   COPYING and README and quite clear GMGPL notices in most (almost
   all) of the file headers.

 - GtkAda 2.2.1 is up to this day linked from Freshmeat with a GMGPL
   license notice and is actually served from libre2.adacore.com.

 - Newer versions of GtkAda are distributed from libre2.adacore.com by
   two protocols: CVS and http(s). The web site does not carry a
   license notice. 

 - Per HTTP GtkAda 2.4.0 is distributed standalone and
   carries the same notices in the files as the "other" version from
   old libre. GtkAda 2.8.0 is distributed by HTTP as part of Gnat-GPL
   and is as such included in the GPL license notice of the GNAT-GPL
   distro. The header files have been supposedly stripped of the
   linking exception (I don't remember wether I checked that
   personally or wether I'm relying in other peoples contribution in
   this single item: If it makes a difference, please check yourself).

 - Both versions 2.4.0 and 2.8.0 and all/most of the older versions
   are distributed by CVS with the license notices identical to the
   pre 2005 status when everything was under GMGPL.

 - AdaCore has answered requests by the Debian maintainer, that all
   Software distributed from libre2 is under pure GPL. 

   Despite the fact that they are using the word "downloaded" in
   answers to my own requests rather often, I think it is quite clear
   that the CVS versions are supposed to be included in that: The
   CVS-Server is located at libre.adacore.com.


Clarification Requests to AdaCore
----------------------------------

Since the AdaCore statements on the licensing terms have always been
somewhat fuzzy in the recent past (see the absence oif notices instead
of clear license notices at the sit) and, interestingly enough, the
license notices in the (supposedly) GMGPL and the (supposedly) GPL
versions have been the same, I've been trying to elicit a statement
from AdaCore on this sameness, on the licensing history of GtkAda,
basically to establish

  - when the license switch actually happened, 
  - how GPL source can be distinguished from the GMGPL source
  
in an effort to establish a last GMGPL baseline from which (depending
on the prevailing needs) either

 - existing projects can be maintained with that version

 - existing projects can be migrated to pure GPL in a controlled way.

 - a GMGPL maintenance branch can be forked as ultimate measure when
   necessary.

I've to say that mail exchange with AdaCore was rather slow and
painful. There seemed to be definite reluctance to answer any question
in a meaningful way at all and even after narrowing everything
(preliminary) down to the question of the present licensing status of
GtkAda 2.4.0 AS OBTAINED FROM LIBRE.ACT-EUROPE in the first quarter
of 2005, it took a time to get a definite statement. 


The Result: "Old" GtkAda is also GPL
------------------------------------

What I got, finally, was the following statement:

    "Let me be clearer, the only assumption you can make today for GtkAda downloaded 
     from the libre site in the past is that it is licensed under the GPL."

Apart from the weasel wording like "assumptions" (which still in my
eyes tries to avoid to talk about the actual rights of the licensee /
person in possession of the code and tries to introduce the notion of
assumptions that could be made, as if AdaCore didn't exactly know
which their licensing terms are or were), apart from that I think that
is quite clear: "Don't use any GtkAda version with a linking
exception". Not now and not in the last year. :-(


Opinion
-------

In the past a number of people here have been trying to portrait
AdaCore as hardcore free software activists that only want to free the
user from evil lesser licenses. I've always thought that this is utter
BS and that AdaCore has indeed changed their direction quite a bit
purely out of business considerations (and you're free to draw your
own conclusion on why such a course change (and in rather clandestine
a way) is necessary if the business model is to sell support for free
software).

But retrospectively revoking licenses should not part of any concept
of freedom and I'm hard pressed to imagine such a scenario. I've gone
as far as stating that it is this practice which will dissolve any
meaning of the license notices in the distributions and any concept of
possesion of a certain distribution under a non revokable
license. I'll repeat it. You'd have to check for changes in the
licenses with the distributor every week.

This (the practice of bad license documentation and the (rather
implicit) retrospective revokation, not the GPL per se) damages free
software and just now it's AdaCore doing the deed.

Of course if it comes to talking to potential customers, it's all COTS
from FLOSS and openness. In my book openness starts with creating
clearly understandable, fixed, licensing situations which I can file
and forget after having read them once. AdaCore fails to to that in a
number of ways (license notices on sites, license notices in distro,
non contradictory notices in alignement with the supposed license).

Congratulations to all people still distributing their GtkAda
applications in binary form based on GMGPL sources. Let me repeat:

 "(...) the only assumption you can make today from the libre site in
  the past is that it is licensed under the GPL (...)"

You got pwned. All your base belong to us. 

Regards -- Markus

Re: Answer of Request to AdaCore on licensing Status of GtkAda 2.4.0

[Michael Bode]

M E Leypold <development-2006-8ecbb5cc8aREMOVETHIS@ANDTHATm-e-leypold.de> writes:

> What I got, finally, was the following statement:
>
>     "Let me be clearer, the only assumption you can make today for GtkAda downloaded 
>      from the libre site in the past is that it is licensed under the GPL."
>

At this point I would be seriously worried if I were a paying customer
of Adacore. What if they suddenly change their licenses for the Pro
versions too? What if that already *has* *happened* at some point in
the *future*? You need a whole new English grammar to describe this
"time travelling license".

This is a company I would definitively never buy from.

-- 
Michael Bode

Re: Answer of Request to AdaCore on licensing Status of GtkAda 2.4.0

[Georg Bauhaus]

Michael Bode wrote:
> M E Leypold <development-2006-8ecbb5cc8aREMOVETHIS@ANDTHATm-e-leypold.de> writes:
> 
>> What I got, finally, was the following statement:
>>
>>     "Let me be clearer, the only assumption you can make today for GtkAda downloaded 
>>      from the libre site in the past is that it is licensed under the GPL."
>>
> 
> At this point I would be seriously worried if I were a paying customer
> of Adacore. What if they suddenly change their licenses for the Pro
> versions too?

"What if..." question tend to introduce FUD. I think you will have
no trouble finding quite definite statements that AdaCore,
like other Ada shops, make sure that their customers won't have any
licensing issues.

As Hyman Rosen has said, it will be judges making judgments on the
licensing state of software based on available evidence etc.
This requires a case.
All of this might not appear in c.l.ada. :-)

Re: Answer of Request to AdaCore on licensing Status of GtkAda 2.4.0

[Hyman Rosen]

M E Leypold wrote:
> You got pwned. All your base belong to us.

You are being silly. Whatever you downloaded came with
the standard GPLish or LGPLish license files. The only
thing you're bound by is what those files say. If you
have a downloaded version with the linking exception,
you're fine (barring that strict reading I mentioned).

Re: Answer of Request to AdaCore on licensing Status of GtkAda 2.4.0

[Michael Bode]

Georg Bauhaus <bauhaus@futureapps.de> writes:

> "What if..." question tend to introduce FUD. 

FUD is correct but it is not introduced by questions but by statements
like that quoted in my post.

> I think you will have no trouble finding quite definite statements
> that AdaCore, like other Ada shops, make sure that their customers
> won't have any licensing issues.

I don't have trouble finding definite statetements about non-customer
licenses too. Example:

  This package is distributed under the GPL license, slightly modified
  so that you can create proprietary software with this toolkit. The
  license is actually the same as the GNAT library itself. You should
  also read the Gtk license itself if you intend to do proprietary
  software based on gtk and GtkAda.

The problem is they are fluctuating in time. What makes you so sure
other statements have higher time-invariance?

> As Hyman Rosen has said, it will be judges making judgments on the
> licensing state of software based on available evidence etc.
> This requires a case.

That's right. But I'd like to avoid beeing party in such a case.

> All of this might not appear in c.l.ada. :-)

Maybe on Groklaw ;-)

-- 
Michael Bode

Re: Answer of Request to AdaCore on licensing Status of GtkAda 2.4.0

[Björn Persson]

The answers I have managed to get from Adacore are quite different from
what Ludovic and Markus have reported. These answers are far from crystal
clear, but the best way I can interpret them is as follows:

� There is *not* a policy that license statements in files headers and
readme files shouldn't be believed.
� Their licensing policy hasn't changed in any significant way. (Yeah
right!)
� They think all the license statements in the packages accurately reflect
the license you get.
� When you download the libraries bundled with Gnat GPL, you get a pure
GPL for the whole bundle.
� As for what license you get when you download a library separately: No
comment.

When there are contradicting claims, I'll go for the one that is closest
to the source. That means I'll stick to the license that is stated in each
piece of code, as downloaded directly from Libre (for packages whose
copyright belongs to Adacore). It's not that I think Ludovic or Markus is
lying, but anything posted here by a third party is technically a rumor
(including this post).

So it seems that for most of the libraries I get a pure GPL if I download
a tarball, but GMGPL if I download it from CVS. As there are contradictions
in some tarballs it may be safest to go for CVS; the license statements
seem to be more consistent there.

Here's my email exchange with Adacore:



From: Bj�rn Persson
To: libre
Subject: How can I find out what license I have recieved?
Date: Fri, 30 Jun 2006 23:43:03 +0200

As I'm sure you know, Adacore's license practice has caused a lot of
discussion in comp.lang.ada recently. Representatives of Adacore have been
quoted as saying that one should not believe the license statements in
software packages downloaded from your website. There have been claims that
it is your position that the only way to find out the license is to ask you
directly, but I can't really believe that you want to answer an email each
and every time someone downloads a file. There is also the problem that
people aren't aware of this policy. Apparently hardly anyone in comp.lang.ada
knew about it until a few days ago, and I'm sure there are lots of people who
don't read comp.lang.ada regularly and still believe that the licenses you
publish are actually the licenses you provide. There must be a better way to
communicate your licenses.

I ask you to publish an official declaration where you state which license
statements should be trusted and which ones should not be trusted. It should
not only cover the current situation but also describe how to know what
license was used at any particular time in the past. I and lots of other
people have software that we have downloaded from your server at various
points in time, and we have believed the license statements that were
included in the packages we dowloaded. We need some way of finding out what
license we really recieved for which package at the time.

Then there's of course one problem: If some things you publish are not meant
to be believed, why should we believe the declaration where you tell us what
to believe and what not to believe? But the approach with direct questions
has exactly the same problem: Why should we believe what you say in emails or
on the phone if we shouldn't believe your website? One thing you could do to
show that this particular declaration should really be trusted is to sign it
with PGP. Putting it on an HTTPS server could also raise the trust level
somewhat.

Regards,
Bj�rn Persson



From: Cyrille Comar
Organization: AdaCore
To: Bj�rn Persson
CC: libre
Subject: Re: How can I find out what license I have recieved?
Date: Sat, 01 Jul 2006 02:30:23 +0200

Bj�rn Persson wrote:

Hello Bjorn,

 > As I'm sure you know, Adacore's license practice has caused a lot of
 > discussion in comp.lang.ada recently.

Yes, it seems that there is a lot of confusion on c.l.a. It is not the
first time and most probably not the last one either ;-)

 > Representatives of Adacore have been
 > quoted as saying that one should not believe the license statements in
 > software packages downloaded from your website. There have been claims that
 > it is your position that the only way to find out the license is to ask you
 > directly, but I can't really believe that you want to answer an email each
 > and every time someone downloads a file.

People are allowed to say whatever they like on c.l.a, nothing we can do
about this. That may have been a reference to a discussion about the
legal value of copyright statements appearing at the top of source files
   that you download over internet in general. I am not a lawyer, not
even an expert myself so I won't make any comment on that specific issue
but there is nothing specific nor different about the libre site in that
respect.

 > There is also the problem that  people aren't aware of this policy.

There is no such policy. We believe the libre site is and has always
been very clear on its intended public (the free-software Ada developers
community) and that the packages available for download contain the
proper GPL statements in the format advised by the FSF. If it is not the
case, please let us know, this is most probably an error and we will be
happy to fix it.

May I ask you to review the main page of the libre site and in
particular the section called "Which version of GNAT is right for you?"
and let me know if it leaves any room for uncertainty?

Very Sincerely,

Cyrille



From: Bj�rn Persson
To: Cyrille Comar
Cc: libre
Subject: Re: How can I find out what license I have recieved?
Date: Thu, 6 Jul 2006 16:59:38 +0200

Cyrille Comar wrote:

 > Yes, it seems that there is a lot of confusion on c.l.a. It is not the
 > first time and most probably not the last one either ;-)

I could mention several things Adacore could have done to reduce the
confusion � or FUD, as some people would call it.

 > People are allowed to say whatever they like on c.l.a, nothing we can do
 > about this.

Of course you can: If false statements are posted you can reply with
corrections. It should be in your company's interest to stop misconceptions
from spreading.

 > That may have been a reference to a discussion about the
 > legal value of copyright statements appearing at the top of source files
 >   that you download over internet in general. I am not a lawyer, not
 > even an expert myself so I won't make any comment on that specific issue
 > but there is nothing specific nor different about the libre site in that
 > respect.
 >
 > > There is also the problem that  people aren't aware of this policy.
 >
 > There is no such policy.

So you mean we should indeed believe the license statements in file headers
and readme files in software downloaded directly from the Libre site? Both
what we download now and what we have downloaded in the past?

Ludovic Brenta has reported that Robert Dewar and Arnaud Charlet have told him
that all software downloaded from AdaCore is pure GPL, no matter what the
headers say. Are you implying that's false?

 > We believe the libre site is and has always
 > been very clear on its intended public (the free-software Ada developers
 > community) and that the packages available for download contain the
 > proper GPL statements in the format advised by the FSF. If it is not the
 > case, please let us know, this is most probably an error and we will be
 > happy to fix it.
 >
 > May I ask you to review the main page of the libre site and in
 > particular the section called "Which version of GNAT is right for you?"
 > and let me know if it leaves any room for uncertainty?

OK, now that I've figured out how to get through the broken firewall I've had
another look. https://libre2.adacore.com/dynamic/gnat_faq.html makes the
licenses of the different editions of Gnat very clear, but it says nothing
about the licenses of ASIS, GTKada, AWS, XML/Ada, GLADE, PolyORB, Aunit or
Florist. The only license statements I have found for those libraries are in
the file headers and readme files, and there are several contradictions
there:

First there's the discrepancy between CVS and the tarballs. When I browse the
CVS repository, AWS, GLADE, GTKada, PolyORB and XML/Ada all have the linking
exception of the GMGPL in the file headers (or at least large parts of them
have it), but in the tarballs the linking exception has been removed from
most files. Apparently I get the GMGPL if I check the code out from CVS, but
if I download tarballs I get the pure GPL � although with some of the
tarballs the situation is murky, as detailed below. Ludovic Brenta says
Robert Dewar and Arnaud Charlet say that code downloaded from the CVS server
is pure GPL. Should I believe Ludovic or should I believe the file headers?
I'm doing neither; I'm asking Adacore: What is the license?

Then there are contradictions within some of the tarballs too:

AWS-gpl-2.2.0/readme.txt and xmlada-gpl-2.2.0/README say AWS and XML/Ada are
distributed under the GMGPL, and AWS-gpl-2.2.0/include/readme.txt lists the
licenses of several components as GMGPL, but in all these cases � except for
two Java files in AWS � the linking exception of the GMGPL has been removed
from the file headers, so that the headers say the license is pure GPL. How
do I know which license is right?

The linking exception has been removed from all the file headers of PolyORB
except for one C file, but the README, INSTALL, NEWS and FEATURES files
contain the linking exception with a wording as if those files were
themselves source code. I don't know whether that means the exception applies
to all of PolyORB or what else "this unit" means in those files.

Additionally, the documentation for AWS and PolyORB contains a whole lot of
source code that has the linking exception, even though it's missing from the
corresponding files in the "src" subdirectories. Again, how do I know which
license is right?

Besides, there is no copy of the GPL to be found in the ASIS or AWS packages,
even though the file headers say I should have recieved one.

Bj�rn Persson



From: Cyrille Comar
Organization: AdaCore
To: Bj�rn Persson
CC: libre
Subject: Re: How can I find out what license I have recieved?
Date: Thu, 13 Jul 2006 17:42:36 +0200

Bj�rn Persson wrote:
 >> People are allowed to say whatever they like on c.l.a, nothing we can do
 >> about this.
 >
 > Of course you can: If false statements are posted you can reply with
 > corrections.

Probably... although doing this consistently would require huge
resources. We prefer to devote our resources to make the technology
progress.

 > Ludovic Brenta has reported that Robert Dewar and Arnaud Charlet have told him
 > that all software downloaded from AdaCore is pure GPL, no matter what the
 > headers say. Are you implying that's false?

I am not implying anything. I am just stating that our policy has not
changed in any significant way: the Libre site has always been dedicated
to Free software developers and has always been advertised this way very
clearly. It has always been AdaCore's position that anything coming from
the libre site was not suitable for non-Libre development. This is why
the site was named that way and why the landing page takes some real
estate for explaining this. We will consider clarifying yet more.

 >> May I ask you to review the main page of the libre site and in
 >> particular the section called "Which version of GNAT is right for you?"
 >> and let me know if it leaves any room for uncertainty?
 >
 > OK, now that I've figured out how to get through the broken firewall I've had
 > another look. https://libre2.adacore.com/dynamic/gnat_faq.html makes the
 > licenses of the different editions of Gnat very clear, but it says nothing
 > about the licenses of ASIS, GTKada, AWS, XML/Ada, GLADE, PolyORB, Aunit or
 > Florist.

Well, all those packages are distributed as part of a GNAT edition. This
is in the context of those editions that we provide explicit license
statements.

 > Besides, there is no copy of the GPL to be found in the ASIS or AWS packages,
 > even though the file headers say I should have recieved one.

Ok, this is helpful, we will fix those packages.
Thank you for your feedback,

Cyrille



-- 
Bj�rn Persson                              PGP key A88682FD
                    omb jor ers @sv ge.
                    r o.b n.p son eri nu

Re: Answer of Request to AdaCore on licensing Status of GtkAda 2.4.0

[michael bode]

=?windows-1252?Q?Bj=F6rn_Persson?= <spam-away@nowhere.nil> writes:

> The answers I have managed to get from Adacore are quite different from
> what Ludovic and Markus have reported. These answers are far from crystal
> clear, but the best way I can interpret them is as follows:
> 
> · There is *not* a policy that license statements in files headers and
> readme files shouldn't be believed.
> · Their licensing policy hasn't changed in any significant way. (Yeah
> right!)
> · They think all the license statements in the packages accurately reflect
> the license you get.

This sounds far better than anything else I've heard here.

> When you download the libraries bundled with Gnat GPL, you get a pure
> GPL for the whole bundle.

Ok.

> As for what license you get when you download a library separately: No
> comment.

They could really be a bit clearer in this section.

[Quote from email answer from Adacore staffer] (I hope no one thinks
this is quoted out of context)

> People are allowed to say whatever they like on c.l.a, nothing we can do
> about this. That may have been a reference to a discussion about the
> legal value of copyright statements appearing at the top of source files
>    that you download over internet in general. I am not a lawyer, not
> even an expert myself so I won't make any comment on that specific issue
> but there is nothing specific nor different about the libre site in that
> respect.

So if it works for other FOSS projects (license-in-header) it works for
Adacore too. Good.

[Quote from 2nd email answer from Adacore staffer]
(I hope no one thinks this is quoted out of context)

> I am not implying anything. I am just stating that our policy has not
> changed in any significant way: the Libre site has always been dedicated
> to Free software developers and has always been advertised this way very
> clearly. It has always been AdaCore's position that anything coming from
> the libre site was not suitable for non-Libre development. This is why
> the site was named that way and why the landing page takes some real
> estate for explaining this. We will consider clarifying yet more.

What Libre site are we talking about? Does "has not changed in any
significant way" imply "since we opened libre2..."? Quote from
http://web.archive.org/web/20050330084052/http://libre.act-europe.fr/
(old Libre site):

  The tools and software that you can download from this site are
  intended for developers of Libre Software, students, teachers and
  hobbyists. If you are planning to develop software in a commercial
  setting (as Libre Software or otherwise) services are available to
  save you time and company money. Contact sales@adacore.com.

So commercial developers can save time and money if they buy a service
contract. That's reasonable. It's not at all clear that they *have* to
do that to develop CSS at all.

http://web.archive.org/web/20050203211706/libre.act-europe.fr/GtkAda/

  This package is distributed under the GPL license, slightly modified
  so that you can create proprietary software with this toolkit. The
  license is actually the same as the GNAT library itself. You should
  also read the Gtk license itself if you intend to do proprietary
  software based on gtk and GtkAda.

http://web.archive.org/web/20050208062827/libre.act-europe.fr/xmlada/

  This library is released under the standard GNAT Modified GNU Public
  License (GMGPL). As usual, it is provided as is, without any
  guarantee or support. We do not recommend using of this package in a
  commercial application. If you are interested in using a supported
  version of this library suitable for commercial applications, please
  contact sales@adacore.com

If nothing has changed since then, fine.

(Of course technically speaking web.archive.org is also only rumor)

If someone from Adacore reads this and it is not too much an effort,
any private or official comments are welcome.

Re: Answer of Request to AdaCore on licensing Status of GtkAda 2.4.0

[Ludovic Brenta]

Björn Persson a écrit :
> The answers I have managed to get from Adacore are quite different from
> what Ludovic and Markus have reported. These answers are far from crystal
> clear, but the best way I can interpret them is as follows:

After reading the rest of your post, it appears to me that the replies
you got were very close to the ones I got; only your interpretation
differs slightly. I agree that the answers we both got left a lot
unanswered.

> · There is *not* a policy that license statements in files headers and
> readme files shouldn't be believed.

My interpretation is: there is no policy at all; they have OTOH
confirmed very explicitly to me that any files downloaded from the CVS
repository is under pure GPL, even though (as I pointed out) most files
contain the linking exception.

Robert Dewar reiterated his statement in an email to me, but also said
he is not allowed to give legal advice, so I conclude that whatever he
said had no legal force either. Catch-22 :)

> · Their licensing policy hasn't changed in any significant way. (Yeah
> right!)

It will be difficult for them to convince us of that. I think they have
always *intended* the libre site and the CVS server for free software
developers *only*, but they have never actively prevented non-free
developers from using their software either, and the letter of the
source files even explicitly allowed such usage. That letter, possibly,
was at odds with AdaCore's spirit.

> · They think all the license statements in the packages accurately reflect
> the license you get.

Neither reply you posted from Cyrille Comar supports that point of
view; he never mentioned the license statements in the packages, in
fact. He only referred to the "landing page" on the libre site.
Furthermore, he alluded to the lack of "suitability" of packages
downloaded from libre for non-libre development, but did not say
whether such development was *allowed* or *forbidden*. It is not clear
to me whether that "suitability" is technical (i.e. not supported) or
legal (i.e. not allowed). A clarification will be most welcome, if they
post one on their web site.

> · When you download the libraries bundled with Gnat GPL, you get a pure
> GPL for the whole bundle.

Yes, that much has always been clear; these source files do not contain
the exception language anymore.

> · As for what license you get when you download a library separately: No
> comment.

Indeed, and this is where the trouble begins. I specifically asked that
question to them and they said it is GPL, even if you download from
their CVS servers and even if the files contain the linking and generic
instantiation exception text.  I did point out to them that the
contradiction was a big source of confusion.

> When there are contradicting claims, I'll go for the one that is closest
> to the source. That means I'll stick to the license that is stated in each
> piece of code, as downloaded directly from Libre (for packages whose
> copyright belongs to Adacore). It's not that I think Ludovic or Markus is
> lying, but anything posted here by a third party is technically a rumor
> (including this post).
>
> So it seems that for most of the libraries I get a pure GPL if I download
> a tarball, but GMGPL if I download it from CVS. As there are contradictions
> in some tarballs it may be safest to go for CVS; the license statements
> seem to be more consistent there.

After what Hyman Rosen said, it would seem that this stance would
probably stand trial in a court, but it contradicts the statements I
received from AdaCore. But IANAL, as usual :/

I have already voiced my position as a Debian developer: I will retain
the GMGPL for existing libraries (those in Sarge) but will switch to
the pure GPL for newer versions (those in Etch), in order to comply
with AdaCore's stated (in email) license. Furthermore, I am removing
the linking and generic instantiation exception text from the source
files in Debian, in order to reduce the potential for confusion; and I
also explain the license change in the copyright file: see the
copyright files in asis [1] and libxmlada2 [2] for the first examples.

[1] http://packages.qa.debian.org/a/asis.html
[2] http://packages.qa.debian.org/libx/libxmlada2.html

-- 
Ludovic Brenta.

Re: Answer of Request to AdaCore on licensing Status of GtkAda 2.4.0

[Simon Clubley]

In article <ptslku4vi4.fsf@hod.lan.m-e-leypold.de>, M E Leypold <development-2006-8ecbb5cc8aREMOVETHIS@ANDTHATm-e-leypold.de> writes:
> 
> Dear All,
> 
> To make it short: A member of the sales departement of AdaCore has
> written a mail to me, which at least tries to stipulate that even the
> older (pre 2005 versions) of GtkAda can only be used under the
> GPL. Which can only mean: The linking exception has been revoked, that
> happening at a unspecified time in the deep past and silently so.
> 

You may wish to ask this person how this position can be in agreement with
the following post from Arnaud Charlet:

http://lists.adacore.com/pipermail/gtkada/2002-February/001204.html

Simon.

-- 
Simon Clubley, clubley@remove_me.eisner.decus.org-Earth.UFP
If Google's motto is "don't be evil", then how did we get Google Groups 2 ?

Re: Answer of Request to AdaCore on licensing Status of GtkAda 2.4.0

[Björn Persson]

Ludovic Brenta wrote:
> Bj�rn Persson a �crit :
>> � They think all the license statements in the packages accurately reflect
>> the license you get.
> 
> Neither reply you posted from Cyrille Comar supports that point of
> view; he never mentioned the license statements in the packages, in
> fact.

Then you missed this sentence: "We believe [...] that the packages 
available for download contain the proper GPL statements in the format 
advised by the FSF."

The sentence after that, "If it is not the case, please let us know, 
this is most probably an error and we will be happy to fix it.", must 
mean that they think the statements are the way they want them to be.

By the way, regarding that mention of "proper GPL statements", it should 
be noted that even GMGPL code contains all the GPL statements that 
pure-GPL code contains.

-- 
Bj�rn Persson                              PGP key A88682FD
                    omb jor ers @sv ge.
                    r o.b n.p son eri nu

Re: Answer of Request to AdaCore on licensing Status of GtkAda 2.4.0

[Steve Whalen]

Ludovic Brenta wrote:
...
> I have already voiced my position as a Debian developer: I will retain
> the GMGPL for existing libraries (those in Sarge) but will switch to
> the pure GPL for newer versions (those in Etch), in order to comply
> with AdaCore's stated (in email) license. Furthermore, I am removing
> the linking and generic instantiation exception text from the source
> files in Debian, in order to reduce the potential for confusion; ....
>

Again, Thanks for all your work on the Debian Ada packages.

One thing I'd request would be for you to clearly distinguish between a
GtkAda or Florist Debian package that is GMGPL and a pure GPL version,
when I'm looking at packages in aptitude or a similar package
management tool.

Whether having two visibly different package names, like maybe
"GPLflorist" vs. "Florist " or GPLgtkAda vs. GtkAda is a good approach
or something else would be better, I dont' know.

I guess part of the request is not to "mix" Debian package dependencies
between GMGPL packages, and the GPL packages.

Is that hard to do? Have you already dealt with this?  I hate to add to
your workload, but if we're going to have two sets of Ada toolkits that
have such radically different permitted uses in Debian, we'll need help
keeping the licenses straight.

Steve

Re: Answer of Request to AdaCore on licensing Status of GtkAda 2.4.0

[Ludovic Brenta]

Steve Whalen a écrit :
> One thing I'd request would be for you to clearly distinguish between a
> GtkAda or Florist Debian package that is GMGPL and a pure GPL version,
> when I'm looking at packages in aptitude or a similar package
> management tool.

I posted a request for comments on 2006-07-14 on this group, where I
asked whether or not I should have parallel versions (GPL and GMGPL) of
libraries. I also said, I think clearly, that I was unwilling to take
this additional burden by myself; therefore, people in need GMGPL of
libraries in Debian were strongly advised to get in touch with me and
offer help.

I have received only three responses (in private), none of which
contained a definitive statement offering help. I conclude that nobody
cares enough to spend the time necessary to maintain the GMGPL
libraries (or else, they think I'll just do the work for them, for
free). So, unless someone steps up, I will simply replace the old GMGPL
versions with the newer pure-GPL ones. Those who need GMGPL libraries
will have to stick with Sarge, or become Debian maintainers themselves
and reintroduce the GMGPL libraries under different names.

I don't need encouragement, I need help.

> Whether having two visibly different package names, like maybe
> "GPLflorist" vs. "Florist " or GPLgtkAda vs. GtkAda is a good approach
> or something else would be better, I dont' know.

In the event that I keep parallel versions, the existing GMGPL packages
will keep their names but the newer packages will receive names
reflecting their version number, as per debian policy. I will consider
adding warning statements in the package descriptions, despite the fact
that all Debian users should know about, and are expected to read the
copyright file shipped with every package.

> I guess part of the request is not to "mix" Debian package dependencies
> between GMGPL packages, and the GPL packages.
>
> Is that hard to do? Have you already dealt with this?  I hate to add to
> your workload, but if we're going to have two sets of Ada toolkits that
> have such radically different permitted uses in Debian, we'll need help
> keeping the licenses straight.

No, keeping track of the licenses is only a small part of the burden;
if need be, we can add pragma License (GPL) statements and let GNAT
help us.

The biggest part of the workload is the transition and upgrade of all
packages (which I am now busy with), long-term maintenance, and
responding to bug reports.

-- 
Ludovic Brenta.

Re: Answer of Request to AdaCore on licensing Status of GtkAda 2.4.0

[michael bode]

"Ludovic Brenta" <ludovic@ludovic-brenta.org> writes:

> I have received only three responses (in private), none of which
> contained a definitive statement offering help. I conclude that nobody
> cares enough to spend the time necessary to maintain the GMGPL
> libraries (or else, they think I'll just do the work for them, for
> free). So, unless someone steps up, I will simply replace the old GMGPL
> versions with the newer pure-GPL ones. Those who need GMGPL libraries
> will have to stick with Sarge, or become Debian maintainers themselves
> and reintroduce the GMGPL libraries under different names.

I will talk about GtkAda only because that's the only lib that I've
used so far and therefore I don't know if there are different
situations for the other libs.

There is now even more confusion than before. 

1. First I read what you heard from Adacore regarding GMGPL:

2.4.0 downloaded before day X: GMGPL
2.4.0 downloaded after day X: GPL
any later version: GPL

2. Then I read what Markus heard from Adacore:

any version: GPL

3. Then I read what Björn Persson heard from Adacore:

versions packaged with GNAT GPL: GPL
versions packaged separate and CVS version: no comment
globally no change in license politics
file headers are correct

Add 2 and 2 and you get 
version from CVS : GMGPL

Depending on what you want to believe, choose from above.

If I were a distribution maintainer, I would do the same thing you
do. If I were interested in an up to date GMGPL version I'd go with
3. and download from CVS. If I were paranoid I'd go with 2. and
couldn't even use 2.4.0 from Sarge for CSS development.

Since there is seemingly not much interest in a 2.4.0 GMGPL version
when the GPL version is at 2.8.x and maybe even such a thing can't
exist (according to 2. above) for me the question simply is: can I
take the GtkAda sources from Sarge and compile them in Etch and roll
my own package (which would then forever stay at 2.4.0)? Probably yes,
I guess or is there any code in 2.4.0 that doesn't compile with gnat
4.1?

The other question would be: would it make more sense to fork an
independent GMGPL GtkAda or join the GNAVI/GWindows project and work
on the GTK+ port of GWindows?

Re: Answer of Request to AdaCore on licensing Status of GtkAda 2.4.0

[Georg Bauhaus]

On Wed, 2006-07-26 at 14:54 +0200, michael bode wrote:


> The other question would be: would it make more sense to fork an
> independent GMGPL GtkAda or join the GNAVI/GWindows project and work
> on the GTK+ port of GWindows?

Notice that GWindows depends on GNAT language extensions,
whereas CLAW, another option, does not.

Re: Answer of Request to AdaCore on licensing Status of GtkAda 2.4.0

[Ludovic Brenta]

michael bode a écrit :
> Since there is seemingly not much interest in a 2.4.0 GMGPL version
> when the GPL version is at 2.8.x and maybe even such a thing can't
> exist (according to 2. above) for me the question simply is: can I
> take the GtkAda sources from Sarge and compile them in Etch and roll
> my own package (which would then forever stay at 2.4.0)? Probably yes,
> I guess or is there any code in 2.4.0 that doesn't compile with gnat
> 4.1?

It is possible. I hinted at that solution in my post: you can become a
Debian developer yourself and maintain GtkAda 2.4 in Etch. I'm not
going to spend the effort if I'm the only one working on it, but I
would welcome co-maintainers.

I don't know if GtkAda 2.4.0 compiles with GCC 4.1. I haven't tried.
But if it doesn't compile, it should be easy to fix as GtkAda does not
depend on the compiler's internals.

> The other question would be: would it make more sense to fork an
> independent GMGPL GtkAda or join the GNAVI/GWindows project and work
> on the GTK+ port of GWindows?

This question has popped up before, but no definitive answer has come
forth. It boils down to: do you care enough that you will spend the
time and effort necessary?  Do not count on others; this is free
software, so scratch your own itch and only then, maybe others will
join you.

-- 
Ludovic Brenta.

Re: Answer of Request to AdaCore on licensing Status of GtkAda 2.4.0

[Dmitry A. Kazakov]

On 26 Jul 2006 14:54:00 +0200, michael bode wrote:

> The other question would be: would it make more sense to fork an
> independent GMGPL GtkAda or join the GNAVI/GWindows project and work
> on the GTK+ port of GWindows?

GtkAda could be reworked from scratch. It must be boring and a lot of work,
but not much complex. GtkAda is quite thin bindings. One could even make
stuff like Gtk_Widget_Record limited controlled and make GTK+ interfaces
Ada interfaces...

-- 
Regards,
Dmitry A. Kazakov
http://www.dmitry-kazakov.de

Re: Answer of Request to AdaCore on licensing Status of GtkAda 2.4.0

[Alex R. Mosteo]

Dmitry A. Kazakov wrote:

> On 26 Jul 2006 14:54:00 +0200, michael bode wrote:
> 
>> The other question would be: would it make more sense to fork an
>> independent GMGPL GtkAda or join the GNAVI/GWindows project and work
>> on the GTK+ port of GWindows?
> 
> GtkAda could be reworked from scratch. It must be boring and a lot of
> work, but not much complex. GtkAda is quite thin bindings. One could even
> make stuff like Gtk_Widget_Record limited controlled and make GTK+
> interfaces Ada interfaces...

Or maybe the people working on the C binding generator of SWIG could use
some help, and start from there. I think that's the really critical Ada
project waiting to happen.

Re: Answer of Request to AdaCore on licensing Status of GtkAda 2.4.0

[Dmitry A. Kazakov]

On Wed, 26 Jul 2006 16:31:24 +0200, Alex R. Mosteo wrote:

> Dmitry A. Kazakov wrote:
> 
>> On 26 Jul 2006 14:54:00 +0200, michael bode wrote:
>> 
>>> The other question would be: would it make more sense to fork an
>>> independent GMGPL GtkAda or join the GNAVI/GWindows project and work
>>> on the GTK+ port of GWindows?
>> 
>> GtkAda could be reworked from scratch. It must be boring and a lot of
>> work, but not much complex. GtkAda is quite thin bindings. One could even
>> make stuff like Gtk_Widget_Record limited controlled and make GTK+
>> interfaces Ada interfaces...
> 
> Or maybe the people working on the C binding generator of SWIG could use
> some help, and start from there. 

Well, GTK+ is a quite weird thing. It tries to do OO in plain C. It is like
flying to the Moon on a bike. (:-)) A generator might break its teeth on
it. GtkAda has difficulties with all this. For example, GTK+ type extension
is not an extension of the corresponding Ada type. GTK+ interface isn't an
Ada interface. GTK+ virtual function isn't a primitive operation etc.

> I think that's the really critical Ada
> project waiting to happen.

SWIG or GTK+ bindings?

-- 
Regards,
Dmitry A. Kazakov
http://www.dmitry-kazakov.de

Re: Answer of Request to AdaCore on licensing Status of GtkAda 2.4.0

[Alex R. Mosteo]

Dmitry A. Kazakov wrote:

> On Wed, 26 Jul 2006 16:31:24 +0200, Alex R. Mosteo wrote:
> 
>> Dmitry A. Kazakov wrote:
>> 
>>> On 26 Jul 2006 14:54:00 +0200, michael bode wrote:
>>> 
>>>> The other question would be: would it make more sense to fork an
>>>> independent GMGPL GtkAda or join the GNAVI/GWindows project and work
>>>> on the GTK+ port of GWindows?
>>> 
>>> GtkAda could be reworked from scratch. It must be boring and a lot of
>>> work, but not much complex. GtkAda is quite thin bindings. One could
>>> even make stuff like Gtk_Widget_Record limited controlled and make GTK+
>>> interfaces Ada interfaces...
>> 
>> Or maybe the people working on the C binding generator of SWIG could use
>> some help, and start from there.
> 
> Well, GTK+ is a quite weird thing. It tries to do OO in plain C. It is
> like flying to the Moon on a bike. (:-)) A generator might break its teeth
> on it. GtkAda has difficulties with all this. For example, GTK+ type
> extension is not an extension of the corresponding Ada type. GTK+
> interface isn't an Ada interface. GTK+ virtual function isn't a primitive
> operation etc.

Yep, no joys in all these things.

>> I think that's the really critical Ada
>> project waiting to happen.
> 
> SWIG or GTK+ bindings?

SWIG

Re: Answer of Request to AdaCore on licensing Status of GtkAda 2.4.0

[Steve Whalen]

Ludovic Brenta wrote:
> Steve Whalen a écrit :
> > One thing I'd request would be for you to clearly distinguish between a
> > GtkAda or Florist Debian package that is GMGPL and a pure GPL version,
> > when I'm looking at packages in aptitude or a similar package
> > management tool.
>
> I posted a request for comments on 2006-07-14 on this group, where I
> asked whether or not I should have parallel versions (GPL and GMGPL) of
> libraries. I also said, I think clearly, that I was unwilling to take
> this additional burden by myself; therefore, people in need GMGPL of
> libraries in Debian were strongly advised to get in touch with me and
> offer help.
>  ...

I phrased my question / request badly.  I was not asking you to do more
work.  I already know there probably would NOT be an Ada compiler in
Debian exept for your work, which I very much appreciate.

> ... I will consider adding warning statements in the package descriptions, ...

I guess I was hoping you could do that without too much trouble.

Perhaps I'm misunderstanding something, but I thought that as long as
anyone has an apt preferences that includes Sarge, and testing / Etch,
both GMGPL and GPL versions of Ada libraries and compilers would be
available in aptitude / dselect / etc.

GtkAda having version 2.4 and 2.8 available could get confusing without
some fairly visible means of distinguishing GMGPL from GPL versions.  I
agree that programmers should know the license of the software they're
downloading and using, but I think it's fairly rare that the "utility"
of a Debian package changes significantly going from one version number
to a higher number, as happens when going from 3.x to 4.x in Debian Ada
compilers or GtaAda from 2.4 to 2.8....

> ... I don't need encouragement, I need help. ...

I wish I had the time to maintain one or more of the GMGPL libraries or
compilers myself, or the budget to assign one or more people to do it,
but I don't.

So we will just have to live with the death of GMGPL Ada (and thus
probably Ada).  That's why I was so upset when AdaCore started down
this path months ago.  I knew this was where we'd end up....

Only AdaCore can keep both GMGPL and GPL versions of the compilers and
libraries it maintains up to date without significant effort and
expense.  

Steve