FSF libgnat

FSF libgnat

[Stanislav Tsekhmistroh]

Hello ev'ryone!

Can anyone tell me, what are the condition for using FSF version of
libgnat? Does linking against it requires providing the sourses of the
"based-on" project?

One more question: which versions of libgnat support tasking and which not?

Re: FSF libgnat

[Ludovic Brenta]

Stanislav Tsekhmistroh <stas@kometalab.com.ua> writes:
> Can anyone tell me, what are the condition for using FSF version of
> libgnat? Does linking against it requires providing the sourses of
> the "based-on" project?

Off the top of my head, the conditions are:

- use the GCC from the FSF that the libgnat is part of; i.e. you
  cannot (without a lot of work) use the GNAT GPL 2005 Edition with
  libgnat from the FSF.  You can choose either GCC 3.4.4 (more stable)
  or GCC 4.0.2 (more features).  Avoid GCC 3.3.6 (Ada is buggy).

- distribute the sources to libgnat, or point people to these sources,
  with your program.  If you modify libgnat, you must publish the
  modified sources of libgnat.

- you can choose the license for your program, you don't have to
  distribute the sources of your program.

> One more question: which versions of libgnat support tasking and
> which not?

All of them, except GCC 3.4.4 on powerpc.

HTH

-- 
Ludovic Brenta.

Re: FSF libgnat

[Björn Persson]

Ludovic Brenta wrote:
> Stanislav Tsekhmistroh <stas@kometalab.com.ua> writes:
>>Can anyone tell me, what are the condition for using FSF version of
>>libgnat?
> 
> - distribute the sources to libgnat, or point people to these sources,
>   with your program.

I *think* you only need to provide the sources if you distribute Libgnat 
as such. If you distribute a program that merely uses Libgnat, then I 
think you don't have to provide any sources, because of the special 
exception.

>   If you modify libgnat, you must publish the
>   modified sources of libgnat.

Unless you only use the modified Libgnat "internally" � within your 
company or similar.

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

Re: FSF libgnat

[Ludovic Brenta]

Björn Persson <spam-away@nowhere.nil> writes:
> Ludovic Brenta wrote:
>> Stanislav Tsekhmistroh <stas@kometalab.com.ua> writes:
>>>Can anyone tell me, what are the condition for using FSF version of
>>>libgnat?
>> - distribute the sources to libgnat, or point people to these
>>   sources, with your program.
>
> I *think* you only need to provide the sources if you distribute
> Libgnat as such. If you distribute a program that merely uses
> Libgnat, then I think you don't have to provide any sources, because
> of the special exception.

Linking libgnat into your program, whether statically or dynamically,
causes parts of it to be incorporated into your program (and, indeed,
causes your program to be a "work based on" libgnat, in GPL parlance).
Your program does not have to be GPL (by special permission), but
libgnat keeps its license, and that license requires that you provide
or offer to provide the sources of libgnat.

>>   If you modify libgnat, you must publish the modified sources of
>>   libgnat.

I'd like to rephrase this: you don't have to publish anything; but if
you distribute your program in binary form to person A, then you must
also give the source code of libgnat to person A, even if you modified
this source code.  You don't have to post the modified sources to a
web site if you don't want to.

> Unless you only use the modified Libgnat "internally" within your
> company or similar.

Correct.  The license only applies if you distribute your program.
But I think that the OP was interested in conditions that apply when
distributing his program.

-- 
Ludovic Brenta.

Re: FSF libgnat

[Stefan Bellon]

Ludovic Brenta wrote:

> >>   If you modify libgnat, you must publish the modified sources of
> >>   libgnat.
> 
> I'd like to rephrase this: you don't have to publish anything; but if
> you distribute your program in binary form to person A, then you must
> also give the source code of libgnat to person A, even if you modified
> this source code.

I'd like to rephrase it again: This only applies if person A is outside
your company. There's an "internal distribution" clause. You only have
to provide the source if you made it "public". Distributing it
"internally" is ok:

http://www.gnu.org/licenses/gpl-faq.html#GPLRequireSourcePostedPublic

-- 
Dipl.-Inf. Stefan Bellon
Bauhaus Software Technologies | TTI GmbH TGZ Softwareanalysen c/o ISTE
Tel.: +49 711 78 16 221       | Universitätsstraße 38
Fax.: +49 711 78 16 380       | 70569 Stuttgart

Re: FSF libgnat

[Björn Persson]

Ludovic Brenta wrote:
> Linking libgnat into your program, whether statically or dynamically,
> causes parts of it to be incorporated into your program (and, indeed,
> causes your program to be a "work based on" libgnat, in GPL parlance).
> Your program does not have to be GPL (by special permission), but
> libgnat keeps its license, and that license requires that you provide
> or offer to provide the sources of libgnat.

In other words, you're saying that different parts of the same 
executable file will be covered by different licenses. Now *that* is messy!

Let's look at the exception:

"As a special exception, if other files instantiate generics from this 
unit, or you link this unit with other files to produce an executable, 
this unit does not by itself cause the resulting executable to be 
covered by the GNU General Public License. This exception does not 
however invalidate any other reasons why the executable file might be 
covered by the GNU Public License."

The wording seems to assume that the GPL would either cover the whole 
executable or none of it, but according to you it should be read as 
"...does not cause the *entire* executable to be covered by the GPL – 
just parts of it."

So what happens when GMGPL code and code with a typical proprietary 
license is combined into a single file? The GMGPL says users are allowed 
to redistribute the file. The proprietary license says they're not. Are 
they only allowed to distribute parts of it? How do they know which 
parts? And who would want assorted shards of machine code anyway?

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

Re: FSF libgnat

[Jeffrey R. Carter]

Björn Persson wrote:
> 
> In other words, you're saying that different parts of the same 
> executable file will be covered by different licenses. Now *that* is messy!

It would be if it were correct, but it is not.

> Let's look at the exception:
> 
> "As a special exception, if other files instantiate generics from this 
> unit, or you link this unit with other files to produce an executable, 
> this unit does not by itself cause the resulting executable to be 
> covered by the GNU General Public License. This exception does not 
> however invalidate any other reasons why the executable file might be 
> covered by the GNU Public License."
> 
> The wording seems to assume that the GPL would either cover the whole 
> executable or none of it, but according to you it should be read as 
> "...does not cause the *entire* executable to be covered by the GPL – 
> just parts of it."

Either the executable falls under the GPL, or it doesn't. If it doesn't, you 
don't have to distribute any source code. Since the GMGPL doesn't cause the 
executable to fall under the GPL, it follows that the use of GMGPL source does 
not require the distribution of source code.

-- 
Jeff Carter
"I would never want to belong to any club that
would have someone like me for a member."
Annie Hall
41

Re: FSF libgnat

[Samuel Tardieu]

>>>>> "Bj�rn" == Bj�rn Persson <spam-away@nowhere.nil> writes:

Bj�rn> In other words, you're saying that different parts of the same
Bj�rn> executable file will be covered by different licenses. Now
Bj�rn> *that* is messy!

It is not messy, it is quite a common situation. When you develop
proprietary software and use off-the-shelf components, each one comes
with its own license. When you develop for Microsoft Windows using
Microsoft toolkits for example, those toolkits have their license
which is likely to differ from the one you choose for your own code.

  Sam
-- 
Samuel Tardieu -- sam@rfc1149.net -- http://www.rfc1149.net/sam

Re: FSF libgnat

[Björn Persson]

Samuel Tardieu wrote:
> When you develop for Microsoft Windows using
> Microsoft toolkits for example, those toolkits have their license
> which is likely to differ from the one you choose for your own code.

I don't have any of Microsoft's toolkits so I can't look at their 
licenses. What license terms do they impose on the binaries you make and 
distribute?

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

Re: FSF libgnat

[Samuel Tardieu]

>>>>> "Bj�rn" == Bj�rn Persson <spam-away@nowhere.nil> writes:

Bj�rn> I don't have any of Microsoft's toolkits so I can't look at
Bj�rn> their licenses. What license terms do they impose on the
Bj�rn> binaries you make and distribute?

I am not a Microsoft user but a quick look at

  http://msdn.microsoft.com/visualc/vctoolkit2003/eula.aspx

reads, for example: (best parts reproduced but others are interesting too)

  "3.1 If you are authorized and choose to redistribute Sample Code or
  Redistributable Code (collectively, the "Redistributables") as
  described in Section 2, you agree:
  [...]
  (iii) to distribute the Licensee Software containing the
  Redistributables pursuant to an end user license agreement (which
  may be "break-the-seal", "click-wrap" or signed), with terms no less
  protective than those contained in this EULA;
  [...]
  (v) to display your own valid copyright notice which shall be
  sufficient to protect Microsoft's copyright in the Software;
  [...]
  (vii) to indemnify, hold harmless, and defend Microsoft from and
  against any claims or lawsuits, including attorney's fees, that
  arise or result from the use or distribution of the Licensee
  Software;"

In particular, (iii) imposes heavy constraints on the license you can
choose for your own software. For example, you may not authorize
decompilation of your software, even if you indicate the addresses
corresponding to your code (to prevent Microsoft code from being
decompiled), because your whole executable must have terms as least as
protective as the Microsoft components you use.

  Sam
-- 
Samuel Tardieu -- sam@rfc1149.net -- http://www.rfc1149.net/sam

Re: FSF libgnat

[Randy Brukardt]

"Samuel Tardieu" <sam@rfc1149.net> wrote in message
news:87slup5l30.fsf@willow.rfc1149.net...
...
> In particular, (iii) imposes heavy constraints on the license you can
> choose for your own software. For example, you may not authorize
> decompilation of your software, even if you indicate the addresses
> corresponding to your code (to prevent Microsoft code from being
> decompiled), because your whole executable must have terms as least as
> protective as the Microsoft components you use.

So far as I know, "redistributables" are various Microsoft DLLs that you can
include with your programs. It's never necessary to include these (most
Windows systems have the ones you need); we don't include any of them with
Janus/Ada for instance, and even using Claw it's unlikely that you'd need
anything not included with all versions of Windows.

I don't recall there being any separate runtime license (although I haven't
read any recent licenses for this information). But I'm not a lawyer...

                      Randy Brukardt.

Re: FSF libgnat

[Björn Persson]

Samuel Tardieu wrote:
> In particular, (iii) imposes heavy constraints on the license you can
> choose for your own software. For example, you may not authorize
> decompilation of your software, even if you indicate the addresses
> corresponding to your code (to prevent Microsoft code from being
> decompiled), because your whole executable must have terms as least as
> protective as the Microsoft components you use.

I see. That's quite similar to how a GPL library dictates the license of 
binaries with that library linked in. I don't see that different 
licenses will apply to different parts of the same file, as Ludovic 
seemed to say happens with the GMGPL. As you write yourself: "your whole 
executable".

In fact, if they *had* allowed you to permit decompilation of your own 
parts of the executable but not of Microsoft's parts, *then* there would 
have been different licenses on different parts of the file. Instead, 
they let you put your own license on the whole file, but require that 
your license contain all of their restrictions.

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

Re: FSF libgnat

[Stanislav Tsekhmistroh]

Tue, 18 Oct 2005 23:33:05 +0200, Ludovic Brenta wrote:

> Björn Persson <spam-away@nowhere.nil> writes:
>> Ludovic Brenta wrote:
>>> Stanislav Tsekhmistroh <stas@kometalab.com.ua> writes:
>>>>Can anyone tell me, what are the condition for using FSF version of
>>>>libgnat?
>>> - distribute the sources to libgnat, or point people to these
>>>   sources, with your program.
>>
>> I *think* you only need to provide the sources if you distribute
>> Libgnat as such. If you distribute a program that merely uses
>> Libgnat, then I think you don't have to provide any sources, because
>> of the special exception.
> 
> Linking libgnat into your program, whether statically or dynamically,
> causes parts of it to be incorporated into your program (and, indeed,
> causes your program to be a "work based on" libgnat, in GPL parlance).
> Your program does not have to be GPL (by special permission), but
> libgnat keeps its license, and that license requires that you provide
> or offer to provide the sources of libgnat.

I.e. conditions for using FSF libgnat (by using I mean dynamic linking
against it) are the same as for using ACT libgnat 3.15p, right?

> 
>>>   If you modify libgnat, you must publish the modified sources of
>>>   libgnat.
It's quite clear :-)

Thank you!

PS: 2 Ludovic :
What are the plans for GNAT in Debian?
What is the preference - to fix/develop/package FSF GNAT or ACT
GNAT 3.15p? And what help is needed most of all?

Re: FSF libgnat

[Björn Persson]

Stanislav Tsekhmistroh wrote:
> I.e. conditions for using FSF libgnat (by using I mean dynamic linking
> against it) are the same as for using ACT libgnat 3.15p, right?

Yes. Both are GMGPL – the Gnat Modified General Public License.

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

GNAT in Debian, heads up (was: FSF libgnat)

[Ludovic Brenta]

Stanislav Tsekhmistroh <stas@kometalab.com.ua> writes:
> PS: 2 Ludovic :
> What are the plans for GNAT in Debian?
> What is the preference - to fix/develop/package FSF GNAT or ACT
> GNAT 3.15p? And what help is needed most of all?

I don't know if you followed the very long thread where a vote took
place, last month, about the next Ada compiler in Debian.  See [1] for
a summary of the votes and the various arguments that were voiced.
There have been other messages since then but I still think that
message captures the gist of the debate.

[1] http://groups.google.com/group/comp.lang.ada/msg/8d1b6d93ee815f87

My plan is to transition to GCC 4.1.  The release branch will be
created by the FSF in the next few weeks, and I anticipate that a
package will appear soon thereafter in Debian's unstable or
experimental distribution.  I will start the transition from there,
probably in November this year.

In the mean time, I'm playing with GNAT 3.15p and breaking things :)

I've modified the build process so that certain parts of the compiler
are exposed in a shared library.  These parts make up the project
manager, and the Ada units that are necessary for the project manager.
The library, which I called "libgnatprj", allows to build
project-aware tools by reusing GNAT's internal code.  One such tool is
AdaBrowse, which gave me the idea in the first place.

A second change which I'm experimenting with is to change the
GNAT_Version_String (declared in gnatvsn.ads, and displayed by
"gnatmake -v") to reflect the Debian version number.  My intention is
to allow ASIS and GLADE to use this shared version string, instead of
providing their own (they both contain a copy of parts of GNAT, and
I'd like to stop that so that they can benefit from bug fixes made in
GNAT).  Another reason for doing this is that the GCC packages
(gnat-3.4, gnat-4.0) also do that.

I've just discovered that changing the GNAT_Version_String with every
Debian release (e.g. "Debian 3.15p-14") was a bad idea, because GNAT
keeps the CRC of the source (gnatvsn.ads) in the corresponding library
information file (gnatvsn.ali).  And if the CRC changes, then
everything that depends on gnatvsn.ads must be recompiled.  The latest
upload is currently gnat 3.15p-16 and it breaks ASIS, GLADE and GPS
(they can still run, but GLADE and GPS fail to build from source).

A second problem I'll tackle in the short term is that libgnatprj
contains some files under GMGPL (e.g. gnatvsn.ads) and other files
under GPL (e.g. the project manager).  I'll want to split that in two
libraries.

I want to port these changes to gnat-4.1 when I start doing the
transition.

-- 
Ludovic Brenta.

Re: GNAT in Debian, heads up

[Ludovic Brenta]

I would like to add:

Ludovic Brenta <ludovic@ludovic-brenta.org> writes:
> Stanislav Tsekhmistroh <stas@kometalab.com.ua> writes:
>> What is the preference - to fix/develop/package FSF GNAT or ACT
>> GNAT 3.15p?

It depends on your goals.  If you just want to write Ada software, I
recommend using Debian Sarge and gnat 3.15p-12.  You will have the
full complement of tools and libraries available to you, and no
"moving target" problem for another 14 months at least.  If you want
to work on the Debian packages for the next release of Debian, you are
most welcome; see below.

>> And what help is needed most of all?

Since I'm planning to start transitioning to GCC 4.1 in a month or so,
now would be a good time to learn the internals of how the gcc-4.0 and
gcc-snapshot packages are built.  The build scripts are fairly
complex, because they have to take care of 8 languages, 12 platforms,
and hundreds of patches.  Later, I will need help with:

* porting relevant patches to gnat-4.1 (from gnat): symbolic tracebacks,
  bug fixes, libgnatprj, etc;

* supporting biarch where it makes sense (especially powerpc64 and
  amd64)

* building gnat-gps 3.0

* porting asis and glade to gnat-4.1

* rebuilding all other packages with the new compiler

in that order.  This will take several months to achieve.

-- 
Ludovic Brenta.

Re: GNAT in Debian, heads up

[Stanislav Tsekhmistroh]

Wed, 19 Oct 2005 20:54:58 +0200, Ludovic Brenta wrote:

> It depends on your goals.  If you just want to write Ada software, I
> recommend using Debian Sarge and gnat 3.15p-12.  You will have the
> full complement of tools and libraries available to you, and no
> "moving target" problem for another 14 months at least.  If you want
> to work on the Debian packages for the next release of Debian, you are
> most welcome; see below.

Since I am relatively new to Ada, my intention is to keep investigating
GNAT and using it in my new projects. (Because I have seen all important
advantages of Ada in my field: my activity is industrial automation.)
While I keep moving on, I'll dig the GNAT sources and try to be usefull.

The situation is like that:
I understand that I need GNAT, but since the ACT rule the GNAT
(at least by this moment :-)), and they don't even stoop to answer my
email messages, I don't see other way than to take what we have and make
all my best. Btw, my company is precisely in the category that "is out of
the ACT scope".

> 
>>> And what help is needed most of all?
> 
> Since I'm planning to start transitioning to GCC 4.1 in a month or so,
> now would be a good time to learn the internals of how the gcc-4.0 and
> gcc-snapshot packages are built.  The build scripts are fairly
> complex, because they have to take care of 8 languages, 12 platforms,
> and hundreds of patches.  Later, I will need help with:
> 
> * porting relevant patches to gnat-4.1 (from gnat): symbolic tracebacks,
>   bug fixes, libgnatprj, etc;
> 
> * supporting biarch where it makes sense (especially powerpc64 and
>   amd64)
> 
> * building gnat-gps 3.0
> 
> * porting asis and glade to gnat-4.1
> 
> * rebuilding all other packages with the new compiler
> 
> in that order.  This will take several months to achieve.

Re: GNAT in Debian, heads up

[Marc A. Criley]

Stanislav Tsekhmistroh wrote:

> The situation is like that:
> I understand that I need GNAT, but since the ACT rule the GNAT
> (at least by this moment :-)), and they don't even stoop to answer my
> email messages, I don't see other way than to take what we have and make
> all my best. Btw, my company is precisely in the category that "is out of
> the ACT scope".

Well, AdaCore is a business, and their business model centers on 
providing high-end support to their paying customers.  (And my past 
experience as a supported customer has been that their support is indeed 
among the industry's best.)  So, if you're requesting technical 
information or support, and you're not a paying customer, you're just 
not going to get much attention.  That said, they do accept bug reports 
from anyone and will pay attention to them, although again, if it's not 
a bug affecting any of their paying customers it likely won't be high on 
the priority list.  (Although I once reported a bug/deficiency while not 
a paying customer and was surprised to almost immediately get questioned 
about it, so bug reports from unsupported users _are_ reviewed.)

Anywho...

There's tons of expertise using GNAT here and in the comp.lang.ada 
archives just waiting to be tapped.

-- Marc A. Criley
-- McKae Technologies
-- www.mckae.com
-- DTraq - XPath In Ada - XML EZ Out