Precisely why can't official FSF GNAT maintainers copy bug fixes in GNAT & its GCC-contained runtime en masse from GNAT GPL Community Edition?

Precisely why can't official FSF GNAT maintainers copy bug fixes in GNAT & its GCC-contained runtime en masse from GNAT GPL Community Edition?

[Dan'l Miller]

Instead of burying this 200 comments down into another thread, this is important enough to deserve its own top-level posting:

On Thursday, May 3, 2018 at 12:00:20 PM UTC-5, Simon Wright wrote: 
> Not actually copyright, since (judging by the 2017 CE) it's assigned to 
> the FSF anyway. It's a licencing issue. 

You do bring up an interesting point:  because AdaCore appears to immediately assign all rights to copy to Free Software Foundation the moment that AdaCore employees invoke the file-save function in their text editor in files comprising GNAT and its portion-of-GCC runtime, it seems that only Free Software Foundation would have legal standing to fully enforce the GPLv3-without-Runtime-Exception in GNAT GPL Community Edition.  As a spectator in the public audience not a party to any such dispute, it would be a fascinating to see years or decades from now whether a maintainer of FSF GNAT who wholesale copied nontrivial portions of GNAT GPL Community Edition into FSF GNAT repositories would evoke the ire of FSF enough (while maintaining FSF's own intellectual property in FSF repositories) for FSF to take legal action effectively against itself.  It is not clear at all precisely which nail AdaCore would hang its “but it is ours all ours” hat on—apparently certainly none of the files that say Copyright Free Software Foundation instead of Copyright AdaCore, Inc. 

Please note that any official maintainer of FSF GNAT acting under authority of FSF would seem free to declare that FSF is evoking its right as owner of the rights to copy GNAT to relicense any portion of GNAT GPL Community Edition to which FSF has been assigned the rights to copy, adjusting the license for that copy from ‘•without• Runtime Exception’ to ‘•with• Runtime Exception’ as GPLv3 seems to overtly permit in section 7 Additional Terms “You may place additional permissions on material, added by you to a covered work, for which you have or can give appropriate copyright permission.”  FSF and its official maintainers deemed to be working on behalf of FSF for FSF-assigned/owned intellectual property [e.g., FSF GNAT] are the “you” therein.  The Runtime Exception would be the “additional permission” therein.  AdaCore's irrevocable assignment to FSF would be the “for which you have or can give appropriate copyright permission” therein; FSF, not AdaCore, effectively owns it at the moment of irrevocable assignment of copyright of that file; that is what assignment means. 

I am not a lawyer and have not passed the bar in any jurisdiction.  I am speaking for only myself in novelty-entertainment value for my own personal enjoyment as a purely-hypothetical/theoretical logic exercise regarding my understanding of plain-meaning reading of English prose.  Do not rely on any of this without consulting a lawyer. 

Re: Precisely why can't official FSF GNAT maintainers copy bug fixes in GNAT & its GCC-contained runtime en masse from GNAT GPL Community Edition?

[Dan'l Miller]

… or brand new features, for that matter?

Precisely why can't official FSF GNAT maintainers copy bug fixes ••and brand new features•• in GNAT & its GCC-contained runtime en masse from GNAT GPL Community Edition?

Re: Precisely why can't official FSF GNAT maintainers copy bug fixes in GNAT & its GCC-contained runtime en masse from GNAT GPL Community Edition?

[Paul Rubin]

"Dan'l Miller" <optikos@verizon.net> writes:
> Precisely why can't official FSF GNAT maintainers copy bug fixes ••and
> brand new features•• in GNAT & its GCC-contained runtime en masse from
> GNAT GPL Community Edition?

I thought the licenses were incompatible.  The FSF runtime has the GPL
runtime exception and the GNAT runtime does not.  Speaking just for
myself, I'd expect the FSF to not accept such a feature if it was aware
of the issue with its licensing.

You also asked what would happen if the FSF accepted the feature without
being aware of the issue, then found out about it later.  That doesn't
seem different than if someone contributed code that they weren't
entitled to contribute (e.g. they wrote it at work, and didn't get the
contribution cleared by their employer).  Again IANAL but ISTM that once
the FSF found out, they'd have to say "oops, despite our efforts to
check out the license of the contribution before we accepted it, it
somehow got past us so we have to withdraw the code since it's not
really ours to release under those terms."  That doesn't sound
conceptually different to me than when someone uploads an unlicensed
song to Youtube and then Youtube has to take it down.

I'd also add that again I'm speaking just for myself, but ISTM that the
FSF is a pro-GPL organization whose goal is for all of the world's
software to be GPL.  It sometimes uses non-GPL licensing such as the
LGPL and the library runtime exception for specific programs as tactical
maneuvers in pursuit of the larger goal, like in chess, where you might
sacrifice a pawn to get closer to checkmating the other player's king.
So I wouldn't expect it to go looking for ways to circumvent Adacore's
choice of the pure GPL.  It *likes* the GPL and is unlikely to see
GNAT's use of the pure GPL as being bad.

Anyway, when you go around trying to subvert someone else's licensing
choices like you seem to be proposing, you're also inviting the rest of
the world to subvert *your* licensing choices.  It's better to stay
clean, I would say.

Re: Precisely why can't official FSF GNAT maintainers copy bug fixes in GNAT & its GCC-contained runtime en masse from GNAT GPL Community Edition?

[Dan'l Miller]

On Thursday, May 3, 2018 at 4:17:30 PM UTC-5, Paul Rubin wrote:
> "Dan'l Miller" writes:
> > Precisely why can't official FSF GNAT maintainers copy bug fixes ••and
> > brand new features•• in GNAT & its GCC-contained runtime en masse from
> > GNAT GPL Community Edition?
> 
> I thought the licenses were incompatible.

They appear to be 100% compatible:  FSF appears to be quite able to override at will AdaCore's removal of the Runtime Exception.  Apparently nothing stops FSF from doing so; just mechanically turn the legal crank of various clauses in GPLv3, so therefore no incompatibility.  I am claiming that ••no one other than•• FSF or an official actor therein (e.g., perhaps an official maintainer; certainly an officer/member of the 501(c)(3)) can re-license files to which FSF is the irrevocable assignee.

>  The FSF runtime has the GPL
> runtime exception and the GNAT runtime does not.  Speaking just for
> myself, I'd expect the FSF to not accept such a feature if it was aware
> of the issue with its licensing.
> 
> You also asked

I asked no such thing.  I made logical statements regarding merely turning the crank of logic and plain-English meaning of English prose.  Statements are not questions.  I did in fact ask a quite different question:  Precisely why [in the license, in case law, in judicial precedent] can't official FSF GNAT maintainers copy bug fixes in GNAT & its GCC-contained runtime en masse from GNAT GPL Community Edition?

> what would happen if the FSF accepted the feature without
> being aware of the issue, then found out about it later.  That doesn't
> seem different than if someone contributed code that they weren't
> entitled to contribute (e.g. they wrote it at work, and didn't get the
> contribution cleared by their employer).  Again IANAL but ISTM that once
> the FSF found out, they'd have to say "oops, despite our efforts to
> check out the license of the contribution before we accepted it, it
> somehow got past us so we have to withdraw the code since it's not
> really ours

  But the rights to copy have already been irrevocably •assigned• to FSF, so in fact it really truly is FSF's at some point prior to AdaCore releasing GNAT GPL Community Edition.

> to release under those terms."  That doesn't sound
> conceptually different to me than when someone uploads an unlicensed
> song to Youtube and then Youtube has to take it down.

  The rights to copy the song were not assigned to the uploader.  The rights to copy the song were not assigned to YouTube.  Non sequitur.

> I'd also add that again I'm speaking just for myself, but ISTM that the
> FSF is a pro-GPL organization whose goal is for all of the world's
> software to be GPL.  It sometimes uses non-GPL licensing such as the
> LGPL and the library runtime exception for specific programs as tactical
> maneuvers in pursuit of the larger goal, like in chess, where you might
> sacrifice a pawn to get closer to checkmating the other player's king.

So are the official maintainers of FSF GNAT merely expendable pawns in the game?  Should someone tell them about their apparent status?

> So I wouldn't expect it to go looking for ways to circumvent Adacore's
> choice of the pure GPL.  It *likes* the GPL and is unlikely to see
> GNAT's use of the pure GPL as being bad.
> 
> Anyway, when you

“you” = AdaCore here?

> go around trying to subvert someone else's

“someone else” = FSF here?

> licensing
> choices like you seem to be proposing, you're also inviting the rest of
> the world to subvert *your* licensing choices.  It's better to stay
> clean, I would say.

Remember, wasn't AdaCore the party who removed the FSF's Runtime Exception via a script •after• AdaCore irrevocably assigning the source code to FSF?  Who is doing the subverting of whose license-choice on files for which who owns the right to copy under assignment?  Answering these questions truthfully is quite awkward.

Re: Precisely why can't official FSF GNAT maintainers copy bug fixes in GNAT & its GCC-contained runtime en masse from GNAT GPL Community Edition?

[Simon Wright]

"Dan'l Miller" <optikos@verizon.net> writes:

> … or brand new features, for that matter?
>
> Precisely why can't official FSF GNAT maintainers copy bug fixes ••and
> brand new features•• in GNAT & its GCC-contained runtime en masse from
> GNAT GPL Community Edition?

I haven't read much of your walls of text (where do you find the time?)
but this is confused.

-----------------

(1) Someone (Randy? one of us?) finds a bug in CE & reports it to
AdaCore.

(2a) It's already fixed in their internal repo: they may tell the
reporter.

(2b) It's new. Entered into their tracking system, prioritised, fixed.

(3) It's released to their customers. Repeat from (1) until ...

(4) Feature freeze for CE release.

-----------------

(1) Start of new FSF major release: grand port of AdaCore changes to FSF
repo.

(2) Integration with parallel changes in GCC backend etc

(3) People build for their weird architectures (incl. Darwin! my
personal interest), find & report problems; sometimes they propose
patches, sometimes they get accepted (internal compiler errors and
bootstrap failures get priority)

(3a) I hove no evidence for this, but I'd expect that serious bugs get
fed back to AdaCore's own system.

(3b) Nor have I evidence that new AdaCore work trickles through into the
FSF system. I did report one bug (80888) and submitted a patch; when the
code was eventually patched, an AdaCore engineer had developed an
identical patch. Not bitter, no! so long as the product is improved.

(4) Repeat from (2) until ...

(5) Release (current is 8.1, just yesterday).

(6) Repeat from (2) for minor releases (next will be 8.2).

-----------------

The thing to remember that AdaCore don't have separate people working on
customer, CE, or FSF compilers.

Re: Precisely why can't official FSF GNAT maintainers copy bug fixes in GNAT & its GCC-contained runtime en masse from GNAT GPL Community Edition?

[Simon Wright]

Simon Wright <simon@pushface.org> writes:

> The thing to remember that AdaCore don't have separate people working
> on customer, CE, or FSF compilers.

And they are the official maintainers.

Re: Precisely why can't official FSF GNAT maintainers copy bug fixes in GNAT & its GCC-contained runtime en masse from GNAT GPL Community Edition?

[Paul Rubin]

"Dan'l Miller" <optikos@verizon.net> writes:
> But the rights to copy have already been irrevocably •assigned• to FSF,

Do you know the terms of the assignment contract?

>> Anyway, when you
> “you” = AdaCore here?

No, "you" = Dan'l Miller, proposing ways to circumvent the Adacore
license by switcheroo'ing it with the FSF license that you like better.

> Remember, wasn't AdaCore the party who removed the FSF's Runtime
> Exception via a script •after• AdaCore irrevocably assigning the
> source code to FSF?

The runtime exception still applies to the version that FSF got.

> So are the official maintainers of FSF GNAT merely expendable pawns in
> the game?  Should someone tell them about their apparent status?

I doubt they are in the dark about anything.  The FSF decided that a
certain maneuver was worthwhile, and the maintainers chose to help the
maneuver along.

> Answering these questions truthfully is quite awkward.

I don't see it.  There's an agreement between Adacore and the FSF that
the rest of us aren't parties to.  If it's working smoothly it's not
awkward.  At worst there could be hypothetical awkwardness if they
started to want differing things.  But they don't seem to want that.

Re: Precisely why can't official FSF GNAT maintainers copy bug fixes in GNAT & its GCC-contained runtime en masse from GNAT GPL Community Edition?

[Dan'l Miller]

On Thursday, May 3, 2018 at 4:48:26 PM UTC-5, Simon Wright wrote:
> (3b) Nor have I evidence that new AdaCore work trickles through into the
> FSF system.

Yes, that is the beef that some people are claiming:  AdaCore GNAT work not showing up in FSF GNAT.  I'm glad you concur at some level there.

> The thing to remember that AdaCore don't have separate people working 
> on customer, CE, or FSF compilers. 
> ...
> And they are the official maintainers. 

Yes, one would expect AdaCore GNAT work to show up in FSF GNAT if they are in fact the only official maintainers, wouldn't one?

Re: Precisely why can't official FSF GNAT maintainers copy bug fixes in GNAT & its GCC-contained runtime en masse from GNAT GPL Community Edition?

[Dan'l Miller]

On Thursday, May 3, 2018 at 5:02:52 PM UTC-5, Paul Rubin wrote:
> "Dan'l Miller" writes:
> > But the rights to copy have already been irrevocably •assigned• to FSF,
> 
> Do you know the terms of the assignment contract?

“When a group at NYU developed the GNU Ada Compiler, with funding from the US Air Force, the contract explicitly called for donating the resulting code to the Free Software Foundation.”
https://www.gnu.org/philosophy/university.html

It is only a matter of effort to find that US Air Force contract in the public records of the federal government.

Here is some prior(-to-now) edition of the assignment contract:
http://www.dreamsongs.com/IHE/IHE-110.html

This is the process, specifying the content of the assignment contract:
https://www.gnu.org/prep/maintain/html_node/Copyright-Papers.html

pertinent commentary on it:
https://www.gnu.org/licenses/why-assign.en.html

> >> Anyway, when you
> > “you” = AdaCore here?
> 
> No, "you" = Dan'l Miller,

No, I am not an official maintainer of FSF GNAT.  I speak only of official maintainers of FSF GNAT.  Quit casting aspersions on my personal asparagus.  I am merely a passive observer, a spectator, chewing popcorn, sitting in my seat in the audience.

> > Remember, wasn't AdaCore the party who removed the FSF's Runtime
> > Exception via a script •after• AdaCore irrevocably assigning the
> > source code to FSF?
> 
> The runtime exception still applies to the version that FSF got.

What FSF got was irrevocable assignment of all ownership.  The content of the files doesn't matter to the owner of the legal rights to that content.  As assignee, FSF owns the content of those rights-assigned files no matter what the assigner legitimately put into them as work for hire.

Re: Precisely why can't official FSF GNAT maintainers copy bug fixes in GNAT & its GCC-contained runtime en masse from GNAT GPL Community Edition?

[Simon Wright]

"Dan'l Miller" <optikos@verizon.net> writes:

> On Thursday, May 3, 2018 at 4:48:26 PM UTC-5, Simon Wright wrote:
>> (3b) Nor have I evidence that new AdaCore work trickles through into
>> the FSF system.
>
> Yes, that is the beef that some people are claiming: AdaCore GNAT work
> not showing up in FSF GNAT.  I'm glad you concur at some level there.

It's not that it doesn't show up _ever_, just that it may well not show
up in FSF until the next major release. Depends how significant it is
(and how much work it would be).

>> The thing to remember that AdaCore don't have separate people working 
>> on customer, CE, or FSF compilers. 
>> ...
>> And they are the official maintainers. 
>
> Yes, one would expect AdaCore GNAT work to show up in FSF GNAT if they
> are in fact the only official maintainers, wouldn't one?

I don't see what extra point you're making.

Re: Precisely why can't official FSF GNAT maintainers copy bug fixes in GNAT & its GCC-contained runtime en masse from GNAT GPL Community Edition?

[Brian Drummond]

On Thu, 03 May 2018 12:13:19 -0700, Dan'l Miller wrote:

> Instead of burying this 200 comments down into another thread, this is
> important enough to deserve its own top-level posting:
> 
> On Thursday, May 3, 2018 at 12:00:20 PM UTC-5, Simon Wright wrote:
>> Not actually copyright, since (judging by the 2017 CE) it's assigned to
>> the FSF anyway. It's a licencing issue.
> 
> You do bring up an interesting point:  because AdaCore appears to
> immediately assign all rights to copy to Free Software Foundation the
> moment that AdaCore employees invoke the file-save function in their
> text editor in files comprising GNAT and its portion-of-GCC runtime, it
> seems that only Free Software Foundation would have legal standing to
> fully enforce the GPLv3-without-Runtime-Exception in GNAT GPL Community
> Edition. 

It may very well be that the problems aren't legal but technical.

Having done a little maintenance on ghdl/gcc (updating it from gcc4.4ish 
to gcc4.6-gcc4.8 a few years ago, this seems to have restarted interest 
in an excellent open-source Ada project) I can comment, and support and 
maybe add to Simon's observations.

Gnat Libre edition may be a newer Ada compiler than FSF Gnat (while still 
lagging Gnat Pro by a year or so) but it is based on a fairly old gcc 
base version (partly because of that lag).

And the front-end language interfaces to the gcc middle-end and back-end 
tends to change quite significantly between gcc releases. This broke ghdl 
quite badly at that time, and it took some serious digging (for someone 
not intimate with gcc internals) to find out where, why and what to do 
about it.



So a wholesale copy of Gnat Libre patches from a middle-aged gcc, getting 
close to retirement, into the FSF Gnat source tree based on a leading 
edge gcc is likely to result only in a colossal build failure followed by 
a horde of difficult bugs.



The C layer between Gnat and GCC is quite large (over 50 .c,.h files); 
much larger than ghdl's astonishingly clean layer (a single C file for 
the compiler, a few small ones for the RTS, all the rest is Ada) so I 
could imagine merely maintaining this interface between gcc versions to 
be quite a task in itself, running to stand still, even before porting 
any actual compiler changes.

(Side note 1 : the sole author of ghdl also ported it to use LLVM as a 
third back end. Comparing this with DragonEgg's problems interfacing Gnat 
to LLVM supports the notion that this is not trivial.)

(Side note 2 : he ended up wih a job at Adacore.)

In short, unless you know different, changes probably have to be brought 
over one by one and massaged to work, as "free software" time and effort 
allows.

-- Brian

Re: Precisely why can't official FSF GNAT maintainers copy bug fixes in GNAT & its GCC-contained runtime en masse from GNAT GPL Community Edition?

[Simon Clubley]

On 2018-05-03, Dan'l Miller <optikos@verizon.net> wrote:
> On Thursday, May 3, 2018 at 5:02:52 PM UTC-5, Paul Rubin wrote:
>> The runtime exception still applies to the version that FSF got.
>
> What FSF got was irrevocable assignment of all ownership.  The content of
> the files doesn't matter to the owner of the legal rights to that content. 
> As assignee, FSF owns the content of those rights-assigned files no matter
> what the assigner legitimately put into them as work for hire.

There are two code bases; the AdaCore one and the FSF one.

The FSF assignment controls the FSF code base only.

Someone in AdaCore who is authorised to do so has to decide to push
code from the AdaCore code base into the FSF code base.

At this point, and _only_ at this point, do the FSF have control
rights to this pushed code.

In particular, the FSF cannot just pull code from the AdaCore code base
and re-licence it under the terms of the FSF codebase without the
permission of AdaCore. AdaCore have to push code into the FSF code base.

This is my understanding of the situation and I have never seen
anything which contradicts this.

Simon.

-- 
Simon Clubley, clubley@remove_me.eisner.decus.org-Earth.UFP
Microsoft: Bringing you 1980s technology to a 21st century world

Re: Precisely why can't official FSF GNAT maintainers copy bug fixes in GNAT & its GCC-contained runtime en masse from GNAT GPL Community Edition?

[Dan'l Miller]

On Friday, May 4, 2018 at 2:36:10 AM UTC-5, Simon Wright wrote:
> "Dan'l Miller" writes:
> 
> > On Thursday, May 3, 2018 at 4:48:26 PM UTC-5, Simon Wright wrote:
> >> (3b) Nor have I evidence that new AdaCore work trickles through into
> >> the FSF system.
> >
> > Yes, that is the beef that some people are claiming: AdaCore GNAT work
> > not showing up in FSF GNAT.  I'm glad you concur at some level there.
> 
> It's not that it doesn't show up _ever_, just that it may well not show
> up in FSF until the next major release. Depends how significant it is
> (and how much work it would be).

Everyone, what is the minimum, typical, and maximum latencies that you have observed from the time that an AdaCore GNAT Pro paying customer receives a bug fix or new feature in the GNAT compiler or its runtime to the time that that bug fix or new feature appears in FSF GNAT?

Well, that actually brings up another can of worms to untangle:  Does AdaCore have a beyond-GPL contractual agreement for paying customers of GNAT Pro to refrain from distributing the source code that they receive with GNAT Pro?  Assuming that GNAT Pro is distributed under the GMGPL (and not a proprietary EULA) to paying GNAT Pro customers, under the terms clearly stated in the GMGPL, those paying customers would have a right to have the source code to GNAT Pro compiler and runtime.  Why do no GNAT Pro paying customers exercise their own right to distribute the GNAT Pro free(dom) source code that they receive under GMGPL?  Under the GMGPL (alone), they would seem to have a right to do so.  But paying customers of GNAT Pro have never ever done this.  There must be an extant reason why this redistribution has never ever happened in all these decades.  Do paying customers of GNAT Pro sign a side contract to refrain from distributing the source code to GNAT Pro, overriding the GMGPL?  Precisely which portion of GMGPL would permit such binding side contracts to restrict freedom of distribution of source code of GNAT Pro compiler and runtime?  Or does the GMGPL •categorically• prohibit such side agreements to restrict freedom of distribution of source code of GNAT Pro compiler and runtime?

On Friday, May 4, 2018 at 7:35:42 AM UTC-5, Simon Clubley wrote:
> On 2018-05-03, Dan'l Miller wrote:
> > On Thursday, May 3, 2018 at 5:02:52 PM UTC-5, Paul Rubin wrote:
> >> The runtime exception still applies to the version that FSF got.
> >
> > What FSF got was irrevocable assignment of all ownership.  The content of
> > the files doesn't matter to the owner of the legal rights to that content. 
> > As assignee, FSF owns the content of those rights-assigned files no matter
> > what the assigner legitimately put into them as work for hire.
> 
> There are two code bases; the AdaCore one and the FSF one.

In copyright law, the focus is on the document.  A code base is not a document; a code base is a mere collection of documents.  But a source-code file is a document.  An executable is a document (which is a derived work from its source code).  So this code base or that code base matters not one whit.  What matters is:  does that file within AdaCore say Copyright Free Software Foundation or Copyright AdaCore Technologies, Inc?  For files that were created under the $3 million Air Force contract at New York University, how on earth could those files say Copyright AdaCore Technologies, Inc when NYU assigned all rights to Free Software Foundation decades ago?  Hence, let us assume that the files within AdaCore say Copyright Free Software Foundation in them, as that is what assignment of rights to copy would seem to absolutely require.  Let us assume that the files within AdaCore also specify the GMGPL.  Under the GMGPL, precisely how could Free Software Foundation not be entitled to receive any source code that a paying customer of GNAT Pro receives as a •distribution• of a GMGPLed work under copyright law?  The GMGPL seems to provide for absolutely no latency or delay mechanism, other than the time required to write a recording medium and send it in the postal mail or private courier.  (The other options in the GMGPL are even faster delivery:  download from server or source code is provided along with the executable derived work.)

> The FSF assignment controls the FSF code base only.

Forget about code bases; it is all about the file.  No, the assignment to FSF would control any file (even within AdaCore) that has Copyright Free Software Foundation notice at the top.  Except for a new file, all edits to files at AdaCore dating back to the Ada9X NYU contract would be derivative work of a file whose rights to copy (and distribute in GNAT Pro to paying customers) have already been assigned to FSF decades ago.  So at the moment of distribution of GNAT Pro to paying customers, AdaCore's edits to long-ago-assigned-to-FSF files can only be distributed under the terms of the GMGPL.  Under the terms of the GMGPL, the freedom of the bits is paramount.  What loophole in GMGPL would permit a more-restrictive superdooper license agreement to be laid atop the GMGPL for paying GNAT Pro customers?

> Someone in AdaCore who is authorised to do so has to decide to push
> code from the AdaCore code base into the FSF code base.

Or under the GMGPL (alone), cannot any paying customer of GNAT Pro obtain the source code and redistribute GNAT Pro publicly as both executable derived work and source code?  Once the rights to copy the files comprising the GNAT compiler and its runtime were assigned to FSF long ago, wouldn't the GMGPL be only contract that governs the freedom of those bits downstream from the distribution of GNAT Pro to paying customers?

> At this point, and _only_ at this point, do the FSF have control
> rights to this pushed code.

  I agree minimally that FSF has absolutely no legal basis to forcibly invade AdaCore's walls & servers to acquire that source code.  That is why the GMGPL v2 or v3 says that the source code must be •voluntarily• made available via various mechanisms to remain in compliance with the GPL base license of GMGPL.  The GPL is not about subpoenas forcibly piercing barriers; the GPL is about voluntarily providing downloads from servers and recording media in the mail to remain in compliance with the license.  Precisely which clause & sentence of the GMGPL says that AdaCore can distribute GNAT Pro to paying customers while months or years later “pushing” their derivative work?  (The last I checked, the word “push” doesn't appear in the GMGPL or its base GPL.)  Months or years is a really really long download or a really really long ‘the checks in the mail’-esque mailing of a recording medium.  I don't see such a delay permitted in the letter or spirit of the GMGPL.  Do you?

> In particular, the FSF cannot just pull code from the AdaCore code base
> and re-licence it under the terms of the FSF codebase without the
> permission of AdaCore. AdaCore have to push code into the FSF code base.

On precisely what legal basis could AdaCore assert its rights of ownership under the GMGPL or under the USA's copyright law?  Precisely which clauses & sentences in the GMGPL permit AdaCore any ownership of derivative works of files whose rights to copy were assigned to FSF long ago?  Under precisely what legal basis would FSF as irrevocable assignee not be the owner of files whose rights to copy (and distribute GNAT Pro) were assigned to FSF years ago?

> This is my understanding of the situation and I have never seen
> anything which contradicts this.

Well, if you were to answer the above awkward questions by meticulously quoting chapter & verse from the GMGPL and USA's copyright law, then now you have seen something that would contradict this.  Does the GMGPL and USA copyright law conform closer to my interpretation or to yours?

DISCLAIMER:
I am not a lawyer and have not passed the bar in any jurisdiction.  I am speaking for only myself in novelty-entertainment value for my own personal enjoyment as a purely-hypothetical/theoretical logic exercise regarding my understanding of plain-meaning reading of English prose.  Do not rely on any of this without consulting a lawyer.

Re: Precisely why can't official FSF GNAT maintainers copy bug fixes in GNAT & its GCC-contained runtime en masse from GNAT GPL Community Edition?

[Mark Lorenzen]

On Friday, May 4, 2018 at 4:34:01 PM UTC+2, Dan'l Miller wrote:
> 
> Well, that actually brings up another can of worms to untangle:  Does AdaCore have a beyond-GPL contractual agreement for paying customers of GNAT Pro to refrain from distributing the source code that they receive with GNAT Pro?  Assuming that GNAT Pro is distributed under the GMGPL (and not a proprietary EULA) to paying GNAT Pro customers, under the terms clearly stated in the GMGPL, those paying customers would have a right to have the source code to GNAT Pro compiler and runtime.  Why do no GNAT Pro paying customers exercise their own right to distribute the GNAT Pro free(dom) source code that they receive under GMGPL?  Under the GMGPL (alone), they would seem to have a right to do so.  But paying customers of GNAT Pro have never ever done this.  There must be an extant reason why this redistribution has never ever happened in all these decades.  Do paying customers of GNAT Pro sign a side contract to refrain from distributing the source code to GNAT Pro, overriding the GMGPL?  Precisely which portion of GMGPL would permit such binding side contracts to restrict freedom of distribution of source code of GNAT Pro compiler and runtime?  Or does the GMGPL •categorically• prohibit such side agreements to restrict freedom of distribution of source code of GNAT Pro compiler and runtime?

My employer is (or rather was as the project is now finished) a paying customer. I can see that our copy of GNAT Pro for LEON3 ELF is licensed under GPL 3 with the Runtime Library Exception.

I guess we could redistribute our copy if we wanted to. There is no side contract, hidden agreement or secret handshake. You buy a support contract for a maximum number of developers based on the "honour system" i.e. you don't cheat. After downloading the SW you can use it without any license key, runtime license fee or whatever.

However, we (and probably a lot of other of AdaCore's paying customers) are in the business of developing software for *our* customers - not in the business of redistributing software with the legal risks that it *may* incur if e.g. if it turns out that some parts of GNAT Pro are not licensed under GPL or other legal issues like that. I don't think many customers can see a compelling business case for redistributing GNAT Pro.

Regards,

Mark L

Re: Precisely why can't official FSF GNAT maintainers copy bug fixes in GNAT & its GCC-contained runtime en masse from GNAT GPL Community Edition?

[Simon Wright]

Simon Clubley <clubley@remove_me.eisner.decus.org-Earth.UFP> writes:

> There are two code bases; the AdaCore one and the FSF one.

Pretty sure there's one code base, with a branch that AdaCore use to
support their customers, and another for FSF GCC. The fact that merges
are one-way doesn't negate this.

> The FSF assignment controls the FSF code base only.

When I was a supported AdaCore customer, the source (3.16a1, at least 15
years ago) had (c) FSF with GMGPL extensions.

Further evidence: now that gnatcoll is on github, you can look at the
sources for yourself[1]. In particular, note that COPYING.RUNTIME was
added 7 years ago!

[1] https://github.com/AdaCore/gnatcoll-core

Re: Precisely why can't official FSF GNAT maintainers copy bug fixes in GNAT & its GCC-contained runtime en masse from GNAT GPL Community Edition?

[Jeffrey R. Carter]

On 05/03/2018 11:48 PM, Simon Wright wrote:
> 
> I haven't read much of your walls of text (where do you find the time?)

Maybe he has a c.l.a boilerplate generator.

-- 
Jeff Carter
"Ada has made you lazy and careless. You can write programs in C that
are just as safe by the simple application of super-human diligence."
E. Robert Tisdale
72

Re: Precisely why can't official FSF GNAT maintainers copy bug fixes in GNAT & its GCC-contained runtime en masse from GNAT GPL Community Edition?

[Dan'l Miller]

On Friday, May 4, 2018 at 11:20:30 AM UTC-5, Mark Lorenzen wrote:
> You buy a support contract for a maximum number of developers based on the "honour system"
> i.e. you don't cheat.

I am pretty sure that RMS would not consider the software-freedom rights overtly & meticulously granted under the wording of the GMGPL as “cheating”.

Re: Precisely why can't official FSF GNAT maintainers copy bug fixes in GNAT & its GCC-contained runtime en masse from GNAT GPL Community Edition?

[Dan'l Miller]

On Friday, May 4, 2018 at 11:45:43 AM UTC-5, Jeffrey R. Carter wrote:
> On 05/03/2018 11:48 PM, Simon Wright wrote:
> > 
> > I haven't read much of your walls of text (where do you find the time?)
> 
> Maybe he has a c.l.a boilerplate generator.

I type on the ANSI-modified Dvorak keyboard layout.  That layout really does permit faster typing than QWERTY.

Re: Precisely why can't official FSF GNAT maintainers copy bug fixes in GNAT & its GCC-contained runtime en masse from GNAT GPL Community Edition?

[Simon Clubley]

On 2018-05-04, Dan'l Miller <optikos@verizon.net> wrote:
> On Friday, May 4, 2018 at 11:20:30 AM UTC-5, Mark Lorenzen wrote:
>> You buy a support contract for a maximum number of developers based on the "honour system"
>> i.e. you don't cheat.
>
> I am pretty sure that RMS would not consider the software-freedom rights overtly & meticulously granted under the wording of the GMGPL as ?cheating?.

Note that he said "support contract", not software. That's what is being
sold here if I understand correctly.

Simon.

-- 
Simon Clubley, clubley@remove_me.eisner.decus.org-Earth.UFP
Microsoft: Bringing you 1980s technology to a 21st century world

Re: Precisely why can't official FSF GNAT maintainers copy bug fixes in GNAT & its GCC-contained runtime en masse from GNAT GPL Community Edition?

[Simon Clubley]

On 2018-05-04, Simon Wright <simon@pushface.org> wrote:
> Simon Clubley <clubley@remove_me.eisner.decus.org-Earth.UFP> writes:
>
>> There are two code bases; the AdaCore one and the FSF one.
>
> Pretty sure there's one code base, with a branch that AdaCore use to
> support their customers, and another for FSF GCC. The fact that merges
> are one-way doesn't negate this.
>

Yes, that would be a far better way to put it, thanks.

What I was trying to say was that there are two distinct versions
of the GNAT compiler and that changes are pushed in one direction only.

Simon.

-- 
Simon Clubley, clubley@remove_me.eisner.decus.org-Earth.UFP
Microsoft: Bringing you 1980s technology to a 21st century world

Re: Precisely why can't official FSF GNAT maintainers copy bug fixes in GNAT & its GCC-contained runtime en masse from GNAT GPL Community Edition?

[Simon Clubley]

On 2018-05-04, Dan'l Miller <optikos@verizon.net> wrote:
> On Friday, May 4, 2018 at 2:36:10 AM UTC-5, Simon Wright wrote:
>> 
>> It's not that it doesn't show up _ever_, just that it may well not show
>> up in FSF until the next major release. Depends how significant it is
>> (and how much work it would be).
>
> Everyone, what is the minimum, typical, and maximum latencies that you
> have observed from the time that an AdaCore GNAT Pro paying customer
> receives a bug fix or new feature in the GNAT compiler or its runtime to
> the time that that bug fix or new feature appears in FSF GNAT?
>

One data point:

I have never been able to build a GNAT compiler for VMS Alpha from
the FSF sources even though AdaCore did offer a GNAT Pro version for
VMS.

I don't know if this is due to missing bits in the FSF sources or my
missing out a critical step (VMS has some unique build requirements).
I have successfully built both GNAT compilers and GNAT cross-compilers
in the past however for other targets so I am well familiar with the
overall process.

>
> On Friday, May 4, 2018 at 7:35:42 AM UTC-5, Simon Clubley wrote:
>> In particular, the FSF cannot just pull code from the AdaCore code base
>> and re-licence it under the terms of the FSF codebase without the
>> permission of AdaCore. AdaCore have to push code into the FSF code base.
>
> On precisely what legal basis could AdaCore assert its rights of
> ownership under the GMGPL or under the USA's copyright law?  Precisely
> which clauses & sentences in the GMGPL permit AdaCore any ownership of
> derivative works of files whose rights to copy were assigned to FSF long
> ago?  Under precisely what legal basis would FSF as irrevocable assignee
> not be the owner of files whose rights to copy (and distribute GNAT Pro)
> were assigned to FSF years ago?
>

The question I would ask is whether the FSF were assigned control of
the master copy of GNAT or whether they were assigned control of a
copy of GNAT which is updated at regular intervals from a master
copy controlled and owned by AdaCore.

I've always thought it to be the latter, but I am willing to be
corrected.

I've always thought of it of something like the situation where a vendor
releases software they control under an open licence but then closes
the source for later versions. This is something the vendor is able
to do because the vendor still owns the code even if they released
the earlier versions under an open licence.

All the vendor isn't allowed to do in that case is to revoke the
licence for the earlier versions.

IOW, my impression has been that AdaCore owns the GNAT source code
even though they make a copy of it available to the FSF and that
copy can be controlled by the FSF under the FSF licence.

Simon.

-- 
Simon Clubley, clubley@remove_me.eisner.decus.org-Earth.UFP
Microsoft: Bringing you 1980s technology to a 21st century world

Re: Precisely why can't official FSF GNAT maintainers copy bug fixes in GNAT & its GCC-contained runtime en masse from GNAT GPL Community Edition?

[Dan'l Miller]

On Friday, May 4, 2018 at 12:42:54 PM UTC-5, Simon Clubley wrote:
> On 2018-05-04, Dan'l Miller wrote:
> > On Friday, May 4, 2018 at 2:36:10 AM UTC-5, Simon Wright wrote:
> >> 
> >> It's not that it doesn't show up _ever_, just that it may well not show
> >> up in FSF until the next major release. Depends how significant it is
> >> (and how much work it would be).
> >
> > Everyone, what is the minimum, typical, and maximum latencies that you
> > have observed from the time that an AdaCore GNAT Pro paying customer
> > receives a bug fix or new feature in the GNAT compiler or its runtime to
> > the time that that bug fix or new feature appears in FSF GNAT?
> >
> 
> One data point:
> 
> I have never been able to build a GNAT compiler for VMS Alpha from
> the FSF sources even though AdaCore did offer a GNAT Pro version for
> VMS.
> 
> I don't know if this is due to missing bits in the FSF sources or my
> missing out a critical step (VMS has some unique build requirements).
> I have successfully built both GNAT compilers and GNAT cross-compilers
> in the past however for other targets so I am well familiar with the
> overall process.
> 
> >
> > On Friday, May 4, 2018 at 7:35:42 AM UTC-5, Simon Clubley wrote:
> >> In particular, the FSF cannot just pull code from the AdaCore code base
> >> and re-licence it under the terms of the FSF codebase without the
> >> permission of AdaCore. AdaCore have to push code into the FSF code base.
> >
> > On precisely what legal basis could AdaCore assert its rights of
> > ownership under the GMGPL or under the USA's copyright law?  Precisely
> > which clauses & sentences in the GMGPL permit AdaCore any ownership of
> > derivative works of files whose rights to copy were assigned to FSF long
> > ago?  Under precisely what legal basis would FSF as irrevocable assignee
> > not be the owner of files whose rights to copy (and distribute GNAT Pro)
> > were assigned to FSF years ago?
> >
> 
> The question I would ask is whether the FSF were assigned control of
> the master copy of GNAT or whether they were assigned control of a
> copy of GNAT which is updated at regular intervals from a master
> copy controlled and owned by AdaCore.
> 
> I've always thought it to be the latter, but I am willing to be
> corrected.

I believe only the original Air Force contract with New York University would reveal whether your interpretation is correct.  If that contract stipulates that all files paid for in the original $3 million contract must have their rights to copy immediately assigned to FSF, then it is as I speculate.  If that contract  allowed NYU to own the files prior to a donation step at the point of delivery to the Air Force and/or public, then your interpretation would be correct.

Does anyone have a copy of that contract between the Air Force and NYU?

> I've always thought of it of something like the situation where a vendor

  Isn't the “vendor” (as sole true owner) here FSF as irrevocable assignee of all rights to copy?

  Perhaps the original vendor was NYU if the contract did not require immediate assignment of rights to copy to FSF, but rather only at the out-bound moment of delivery.  As I understand it, NYU's Ada9X compiler under that Air Force contract is a derivative work of NYU's Ada83 compiler source code.

> releases software they control under an open licence but then closes
> the source for later versions. This is something the vendor is able
> to do because the vendor still owns the code even if they released
> the earlier versions under an open licence.

Does AdaCore own the source code?  That is the key question.  If AdaCore owns the source code to GNAT compiler and runtime, wouldn't the file prologues of GNAT's source code in some repository or distribution somewhere anywhere read Copyright AdaCore Technologies, Inc instead?

> All the vendor isn't allowed to do in that case is to revoke the
> licence for the earlier versions.
> 
> IOW, my impression

  The key word there is:  impression.  I doubt that the Air Force contract with NYU for Ada9X mentions the word “impression”.

> has been that AdaCore owns the GNAT source code
> even though they make a copy of it available to the FSF and that
> copy can be controlled by the FSF under the FSF licence.

Impression is not mentioned in the GMGPL the last time that I checked.  I am pretty sure that USA's copyright laws are not predicated on impression.  Your line of reasoning would be more convincing if you were to quote chapter & verse from the GMGPL or United States Code statutory law.

Re: Precisely why can't official FSF GNAT maintainers copy bug fixes in GNAT & its GCC-contained runtime en masse from GNAT GPL Community Edition?

[Dan'l Miller]

On Friday, May 4, 2018 at 12:22:50 PM UTC-5, Simon Clubley wrote:
> On 2018-05-04, Dan'l Miller wrote:
> > On Friday, May 4, 2018 at 11:20:30 AM UTC-5, Mark Lorenzen wrote:
> >> You buy a support contract for a maximum number of developers based on the "honour system"
> >> i.e. you don't cheat.
> >
> > I am pretty sure that RMS would not consider the software-freedom rights overtly & meticulously granted under the wording of the GMGPL as ?cheating?.
> 
> Note that he said "support contract", not software. That's what is being
> sold here if I understand correctly.

Referring to the table:
https://www.adacore.com/gnatpro/comparison

As I understand it, what is being sold as GNAT Pro Enterprise and GNAT Pro Assurance levels is not merely some sort of “honor system” for only “support contract”.  No, what would be being sold is differentiated tangible access to the the following that might be ••entirely omitted•• from GNAT Pro Developer:
1) ability to exclude post-Ada95 language features;
2) ability to exclude post-Ada83 language features;
3) executable for AIX, whose backend targets POWER ISA;
4) executable for Solaris, whose backend targets SPARC & IA32 ISA;
5) executable for some OS, whose backend targets Android natively on ARM ISA;
6) executable for some OS, whose backend targets bare metal on PowerPC and LEON ISAs;
7) executable for some OS, whose backend targets Linux on PowerPC;
8) executable for MacOS(?), whose backend targets iOS on ARM ISA;
9) executable for some OS, whose backend targets Lynx178 on IA32 and PowerPC ISAs;
10) executable for same OS, whose backend targets PikeOS on IA32, ARM, and PowerPC ISAs;
11) executable for some OS, whose backend targets VxWorks on IA32, PowerPC, and ARM.

(a) Are these 11 entirely omitted from GNAT Pro Developer?
(b) Or are GNAT Pro Developer paying customers on the “honor system” to not submit “support contract” requests on these 11 Enterprise-&-Assurance-only platforms?

If (a), then what is being sold (and effectively restricted somehow) isn't merely “support contract”, but rather the GMGPLed derivative-work bits themselves.  (How precisely is that the software freedom that RMS envisions for GMGPLed software?)  If (b), then precisely which portion of the GMGPL says that the paying customer needs to refrain from executing those 11?

Methinks that even Assurance and Enterprise paying customers don't receive all 11 en masse either, but rather only the variants that they overtly chose off the menu at the time of the price quotation.  This compartmentalization works over the long term only if paying customers don't exercise their software-freedom rights in the GMGPL to re-distribute the source code that they have received under the GMGPL's software-freedom terms (or the executables for that matter).  Is an implied pressure to not disobey the honor system tantamount to a beyond-GMGPL license agreement atop the GMGPL?  Must all license agreements that are in effect in reality actually appear in writing?  What happens if a paying customer were to disobey the “honor system”?

DISCLAIMER:
I am not a lawyer and have not passed the bar in any jurisdiction.  I am speaking for only myself in novelty-entertainment value for my own personal enjoyment as a purely-hypothetical/theoretical logic exercise regarding my understanding of plain-meaning reading of English prose.  Do not rely on any of this without consulting a lawyer.

Re: Precisely why can't official FSF GNAT maintainers copy bug fixes in GNAT & its GCC-contained runtime en masse from GNAT GPL Community Edition?

[antispam]

Dan'l Miller <optikos@verizon.net> wrote:
> On Thursday, May 3, 2018 at 5:02:52 PM UTC-5, Paul Rubin wrote:
> > "Dan'l Miller" writes:
> > > But the rights to copy have already been irrevocably ?assigned? to FSF,
> > 
> > Do you know the terms of the assignment contract?
> 
> ?When a group at NYU developed the GNU Ada Compiler, with funding from the US Air Force, the contract explicitly called for donating the resulting code to the Free Software Foundation.?
> https://www.gnu.org/philosophy/university.html
> 
> It is only a matter of effort to find that US Air Force contract in the public records of the federal government.
> 
> Here is some prior(-to-now) edition of the assignment contract:
> http://www.dreamsongs.com/IHE/IHE-110.html
> 
> This is the process, specifying the content of the assignment contract:
> https://www.gnu.org/prep/maintain/html_node/Copyright-Papers.html
> 
> pertinent commentary on it:
> https://www.gnu.org/licenses/why-assign.en.html
> 
> > >> Anyway, when you
> > > ?you? = AdaCore here?
> > 
> > No, "you" = Dan'l Miller,
> 
> No, I am not an official maintainer of FSF GNAT.  I speak only of official maintainers of FSF GNAT.  Quit casting aspersions on my personal asparagus.  I am merely a passive observer, a spectator, chewing popcorn, sitting in my seat in the audience.
> 
> > > Remember, wasn't AdaCore the party who removed the FSF's Runtime
> > > Exception via a script ?after? AdaCore irrevocably assigning the
> > > source code to FSF?
> > 
> > The runtime exception still applies to the version that FSF got.
> 
> What FSF got was irrevocable assignment of all ownership.  The content of the files doesn't matter to the owner of the legal rights to that content.  As assignee, FSF owns the content of those rights-assigned files no matter what the assigner legitimately put into them as work for hire.

I think you have big misunderstanding how copyright works.  FSF
got files at some specific time (in particular due to initial
assignment).  Changes to those file also are covered by
copyright, you can distribute changed files only by agreement
of _all_ copyright owners, in this case Ada Core(additions) and
FSF (base version).  Do not be fooled by copyright statements
in files: copyright exists even if there is no explicit
statement in files.  And changing copyright notices is
extra work (and it interrupts normal software workflow)
so Ada Core used to have wrong copyright notices.
Robert Dewar (who IIUC has law degree) claimed that
copyright notices does not matter, but at some moment
Ada Core decided to put correct license condition into
files.  I guess that they decided to give misleading
statement about ownership as this does not cause
serious problems.

Concerning FSF, I think that they know what Ada Core
is doing and accept that.  I personaly think that
Ada Core is subverting GPL, but apparently FSF
either thinks that what Ada Core is doing is right or
decided to compromise to have a contributor.

Note that without paying customers Ada Core would
get out of business and Ada Core thinks that
it needs licence restrictions to get enough money.
So sticking to GPL spirit as I see it probably
would mean no GNAT in the future or at least
stagnation (adapting GNAT to GCC backend is
a continuous effort that would put significant
load on volunteer maintainers).

-- 
                              Waldek Hebisch

Re: Precisely why can't official FSF GNAT maintainers copy bug fixes in GNAT & its GCC-contained runtime en masse from GNAT GPL Community Edition?

[Dan'l Miller]

On Friday, May 4, 2018 at 2:53:05 PM UTC-5, wrote:
> Dan'l Miller wrote:
> > What FSF got was irrevocable assignment of all ownership.  The content of the files doesn't matter
> > to the owner of the legal rights to that content.  As assignee, FSF owns the content of those
> > rights-assigned files no matter what the assigner legitimately put into them as work for hire.
> 
> I think you have big misunderstanding how copyright works.  FSF
> got files at some specific time (in particular due to initial
> assignment).  Changes to those file also are covered by
> copyright, you can distribute changed files only by agreement
> of _all_ copyright owners,

  I think that you have a big misunderstanding of how irrevocable assignment works.

> in this case Ada Core(additions)

  When these “additions” are edits to files to which the rights to copy have been irrevocably assigned to FSF, then these are derivative works, whose ability to copy are governed by the GMGPL, even for AdaCore to copy, because their or NYU's irrevocable assignment disowned all ownership (hence the word assignment) forever (hence the word irrevocable).

> and
> FSF (base version).

The true base version of the most important files were at NYU, as governed by the Air Force contract for Ada9X Project than in exchange for receiving $3 million reportedly required assigning all rights to copy to FSF.

>  Do not be fooled by copyright statements
> in files: copyright exists

 i.e., copyright exists in 1992 & 1993 at NYU for the most interesting files in GNAT without copyright notices, but now that those files have copyright notices, those notices and their referenced licensing are in full effect, superseding any historical implicit lack of notices in prior editions of the work.  Precisely where in United States statutory law or the ratified treaties does it say otherwise?

> even if there is no explicit
> statement in files.
> ... 
> Concerning FSF, I think that they know what Ada Core
> is doing and accept that.

I would enjoy hearing RMS's statements regarding that topic of being fully aware.

>  I personaly think that
> Ada Core is subverting GPL, but apparently FSF
> either thinks that what Ada Core is doing is right or
> decided to compromise to have a contributor.

I would enjoy hearing RMS's statements regarding that topic of compromising software freedom, for going along to get along.

Re: Precisely why can't official FSF GNAT maintainers copy bug fixes in GNAT & its GCC-contained runtime en masse from GNAT GPL Community Edition?

[Simon Wright]

antispam@math.uni.wroc.pl writes:

>  Do not be fooled by copyright statements
> in files: copyright exists even if there is no explicit
> statement in files.  And changing copyright notices is
> extra work (and it interrupts normal software workflow)
> so Ada Core used to have wrong copyright notices.
> Robert Dewar (who IIUC has law degree) claimed that
> copyright notices does not matter

He did indeed; but FSF statements on how to put your contribution under
GPL say to put the notice in all files.

>    I guess that they decided to give misleading
> statement about ownership as this does not cause
> serious problems.

What misleading statement?

> Concerning FSF, I think that they know what Ada Core
> is doing and accept that.  I personaly think that
> Ada Core is subverting GPL, but apparently FSF
> either thinks that what Ada Core is doing is right or
> decided to compromise to have a contributor.

Many FSF offerings have similar notices; including C library, C++
library, Bison

> Note that without paying customers Ada Core would
> get out of business and Ada Core thinks that
> it needs licence restrictions to get enough money.

People who are developing multimillion's worth of software can usually
be persuaded to buy support for it.

>            (adapting GNAT to GCC backend is
> a continuous effort that would put significant
> load on volunteer maintainers).

Well, that's certainly true.

Re: Precisely why can't official FSF GNAT maintainers copy bug fixes in GNAT & its GCC-contained runtime en masse from GNAT GPL Community Edition?

[Dan'l Miller]

On Friday, May 4, 2018 at 4:47:01 PM UTC-5, Simon Wright wrote:
> antispam@math.uni.wroc.pl writes:
> > Concerning FSF, I think that they know what Ada Core
> > is doing and accept that.  I personaly think that
> > Ada Core is subverting GPL, but apparently FSF
> > either thinks that what Ada Core is doing is right or
> > decided to compromise to have a contributor.
> 
> Many FSF offerings have similar notices; including C library, C++
> library, Bison

I see absolutely no analogue to AdaCore and its GNAT Pro Assurance or Enterprise or Developer editions in the examples that you give:  GCC C library, GCC C++ library, or GNU Bison.  Please provide a URL to a company that sells restricted access to GCC C library, GCC C++ library, or GNU Bison for which source code is unavailable at time of purchase/download to the ••general public••.  I am pretty sure that no such URL exists to such a company for such FSF products anywhere on planet Earth (other than AdaCore regarding C/C++ integration with GNAT Pro Assurance and Enterprise editions).  After the demise of Cygnus Solutions, is AdaCore the only such company on planet Earth selling restricted* access to source code whose rights to copy are assigned to FSF?

* Btw, how precisely is it restricted?  Is it in fact restricted at all?

DISCLAIMER:
I am not a lawyer and have not passed the bar in any jurisdiction.  I am speaking for only myself in novelty-entertainment value for my own personal enjoyment as a purely-hypothetical/theoretical logic exercise regarding my understanding of plain-meaning reading of English prose.  Do not rely on any of this without consulting a lawyer.

Re: Precisely why can't official FSF GNAT maintainers copy bug fixes in GNAT & its GCC-contained runtime en masse from GNAT GPL Community Edition?

[Luke A. Guest]

Simon Clubley <clubley@remove_me.eisner.decus.org-Earth.UFP> wrote:

> Yes, that would be a far better way to put it, thanks.
> 
> What I was trying to say was that there are two distinct versions
> of the GNAT compiler and that changes are pushed in one direction only.

It’s more likely to be three branches with the CODE flowing like:

GNAT Pro -> GNAT GPL -> FSF GNAT

Re: Precisely why can't official FSF GNAT maintainers copy bug fixes in GNAT & its GCC-contained runtime en masse from GNAT GPL Community Edition?

[Luke A. Guest]

Simon Clubley <clubley@remove_me.eisner.decus.org-Earth.UFP> wrote:

> One data point:
> 
> I have never been able to build a GNAT compiler for VMS Alpha from
> the FSF sources even though AdaCore did offer a GNAT Pro version for
> VMS.
>
> I don't know if this is due to missing bits in the FSF sources or my
> missing out a critical step (VMS has some unique build requirements).

It’s missing the bits you need. I’ve even noticed VMS bits having been
remove bed from one version to another.

Luke 

Re: Precisely why can't official FSF GNAT maintainers copy bug fixes in GNAT & its GCC-contained runtime en masse from GNAT GPL Community Edition?

[Dan'l Miller]

On Saturday, May 5, 2018 at 7:44:54 AM UTC-5, Luke A. Guest wrote:
> Simon Clubley <clubley@remove_me.eisner.decus.org-Earth.UFP> wrote:
> 
> > Yes, that would be a far better way to put it, thanks.
> > 
> > What I was trying to say was that there are two distinct versions
> > of the GNAT compiler and that changes are pushed in one direction only.
> 
> It’s more likely to be three branches with the CODE flowing like:
> 
> GNAT Pro -> GNAT GPL -> FSF GNAT

Yes, that is how I suspect that it is as well.

The question is where precisely historically did each of the files is the GNAT-Pro branch originate?  I suspect that even the oldest of these are files whose rights to copy were assigned to FSF by NYU 1992 to 1993, likely predating even the formation of AdaCore (and its twin in France).  My suspicion is that even AdaCore depends on the GMGPL (as any third party would) to license these NYU-authored files in the GNAT Pro branch, especially the oldest ones dating all the way back to NYU.  (Subsequent assignments of rights to copy to FSF •might• have placed newer first-written-by-AdaCore files in the same category too.)

Does anyone have a copy of the Air Force contract with NYU to write the Ada9X Project's compiler?

Re: Precisely why can't official FSF GNAT maintainers copy bug fixes in GNAT & its GCC-contained runtime en masse from GNAT GPL Community Edition?

[Lucretia]

On Saturday, 5 May 2018 15:19:23 UTC+1, Dan'l Miller  wrote:
> On Saturday, May 5, 2018 at 7:44:54 AM UTC-5, Luke A. Guest wrote:
> >
> > GNAT Pro -> GNAT GPL -> FSF GNAT
> 
> Yes, that is how I suspect that it is as well.
> 
> The question is where precisely historically did each of the files is the GNAT-Pro branch originate?  I suspect that even the oldest of these are files whose 

Back in 1995, when I was learning Ada9X, GANT was distributed separately from GCC as it hadn't been accepted into the main tree, AFAIK. As of GCC-2.95.1 which is around EGCS era IIRC, there is no gcc/ada directory. I managed to find these:

http://archive.adaic.com/ase/ase02_02/lang/ada/gnat/3.12p/
http://sunsite.rwth-aachen.de:3080/ftp/pub/mirror/cs.nyu.edu/pub/gnat/3.15p/

You'll find there are copyright notices belonging to FSU as well there.

> rights to copy were assigned to FSF by NYU 1992 to 1993, likely predating even the formation of AdaCore (and its twin in France).  My suspicion is that even 

AdaCore was "spun out of NYU" and I still have no clue how it became a French company.

Re: Precisely why can't official FSF GNAT maintainers copy bug fixes in GNAT & its GCC-contained runtime en masse from GNAT GPL Community Edition?

[Niklas Holsti]

On 18-05-05 18:58 , Lucretia wrote:

> AdaCore was "spun out of NYU" and I still have no clue how it became
> a French company.

In what sense is AdaCore "a French company"?

The AdaCore website is somewhat coy about the company's structure, but 
the Privacy Statement mentions "AdaCore SAS in France, Ada Core 
Technologies Inc. in the United States". So at the very least there is a 
U.S. branch, and I would guess that it is the main branch.

-- 
Niklas Holsti
Tidorum Ltd
niklas holsti tidorum fi
       .      @       .

Re: Precisely why can't official FSF GNAT maintainers copy bug fixes in GNAT & its GCC-contained runtime en masse from GNAT GPL Community Edition?

[Dan'l Miller]

On Saturday, May 5, 2018 at 10:58:56 AM UTC-5, Lucretia wrote:
> On Saturday, 5 May 2018 15:19:23 UTC+1, Dan'l Miller  wrote:
> > On Saturday, May 5, 2018 at 7:44:54 AM UTC-5, Luke A. Guest wrote:
> > >
> > > GNAT Pro -> GNAT GPL -> FSF GNAT
> > 
> > Yes, that is how I suspect that it is as well.
> > 
> > The question is where precisely historically did each of the files is the GNAT-Pro branch originate?  I suspect that even the oldest of these are files whose 
> 
> Back in 1995, when I was learning Ada9X, GANT was distributed separately from GCC as it hadn't been accepted into the main tree, AFAIK. As of GCC-2.95.1 which is around EGCS era IIRC, there is no gcc/ada directory. I managed to find these:
> 
> http://archive.adaic.com/ase/ase02_02/lang/ada/gnat/3.12p/
> http://sunsite.rwth-aachen.de:3080/ftp/pub/mirror/cs.nyu.edu/pub/gnat/3.15p/
> 
> You'll find there are copyright notices belonging to FSU as well there.
> 
> > rights to copy were assigned to FSF by NYU 1992 to 1993, likely predating even the formation of AdaCore (and its twin in France).  My suspicion is that even 
> 
> AdaCore was "spun out of NYU" and I still have no clue how it became a French company.

As shown in a 1996-era website
http://www.cs.scranton.edu/~beidler/Ada/gnat/a-diocst.ads
the assignment of copyright apparently started in 1992, at least for this file in the runtime.

AdaCore's own website states that AdaCore was launched in 1994 after the assignment of the rights to copy were assigned to FSF.

Release 1.83 of GNAT was prior to July 1994, directly from New York University.

Re: Precisely why can't official FSF GNAT maintainers copy bug fixes in GNAT & its GCC-contained runtime en masse from GNAT GPL Community Edition?

[Luke A. Guest]

Niklas Holsti <niklas.holsti@tidorum.invalid> wrote:
> On 18-05-05 18:58 , Lucretia wrote:
> 
>> AdaCore was "spun out of NYU" and I still have no clue how it became
>> a French company.
> 
> In what sense is AdaCore "a French company"?

Both LinkedIn and Glassdoor show AdaCore HQ as Paris not NY.

Adacore French connection

[J-P. Rosen]

Le 05/05/2018 à 17:58, Lucretia a écrit :
> AdaCore was "spun out of NYU" and I still have no clue how it became a French company.
Around 1982, Philippe Kruchten was a professor at ENST (french
engineering school), and had translated the (preliminary) ARM in French,
so he had some connection with the people at NYU. He was invited to a
sabbatical there. I was also professor at ENST, so after 6 months he
pulled me there too.

When we returned to France, the link was established, and we regularly
sent professors for sabbatical and students as interns to NYU. Some of
them were the ones who founded the French branch of AdaCore when they
returned home...

-- 
J-P. Rosen
Adalog
2 rue du Docteur Lombard, 92441 Issy-les-Moulineaux CEDEX
Tel: +33 1 45 29 21 52, Fax: +33 1 45 29 25 00
http://www.adalog.fr

Re: Precisely why can't official FSF GNAT maintainers copy bug fixes in GNAT & its GCC-contained runtime en masse from GNAT GPL Community Edition?

[Simon Wright]

Luke A. Guest <laguest@archeia.com> writes:

> Simon Clubley <clubley@remove_me.eisner.decus.org-Earth.UFP> wrote:
>
>> Yes, that would be a far better way to put it, thanks.
>> 
>> What I was trying to say was that there are two distinct versions
>> of the GNAT compiler and that changes are pushed in one direction only.
>
> It’s more likely to be three branches with the CODE flowing like:
>
> GNAT Pro -> GNAT GPL -> FSF GNAT

Considering the timing, I'd think each GPL release is a bookmark (tag)
in the GNAT Pro repo, somewhat after the Pro=>FSF merge.

Hard to say really about the relative dates: GCC 8.1.0 has just been
released, and there has been no massive activity in the repo since. Only
a few days, though, give it time!

Things do happen between releases, though: for instance, when I started
following GCC 8 for Cortex GNAT RTS, the interface for task creation in
s-tarest.ads was the same as GPL 2017; now it isn't
(Create_Restricted_Task has new parameter Sec_Stack_Address).

Re: Precisely why can't official FSF GNAT maintainers copy bug fixes in GNAT & its GCC-contained runtime en masse from GNAT GPL Community Edition?

[Jacob Sparre Andersen]

Simon Wright wrote:

> Considering the timing, I'd think each GPL release is a bookmark (tag)
> in the GNAT Pro repo, somewhat after the Pro=>FSF merge.

This is also how I understand the process.

Greetings,

Jacob
-- 
"There's a lot of information that was in the programmer's head
 when they wrote the code that isn't reflected in that code"

Re: Precisely why can't official FSF GNAT maintainers copy bug fixes in GNAT & its GCC-contained runtime en masse from GNAT GPL Community Edition?

[Simon Clubley]

On 2018-05-05, Luke A  Guest <laguest@archeia.com> wrote:
> Simon Clubley <clubley@remove_me.eisner.decus.org-Earth.UFP> wrote:
>
>> One data point:
>> 
>> I have never been able to build a GNAT compiler for VMS Alpha from
>> the FSF sources even though AdaCore did offer a GNAT Pro version for
>> VMS.
>>
>> I don't know if this is due to missing bits in the FSF sources or my
>> missing out a critical step (VMS has some unique build requirements).
>
> It?s missing the bits you need. I?ve even noticed VMS bits having been
> remove bed from one version to another.
>

Thank you Luke, that's very useful. It means I now know for sure
there's no point in spending any more time in the future trying
to get it working.

(After a major GCC version release, I sometimes have an attempt at
trying to build GNAT for VMS but so far without success.)

Thanks,

Simon.

-- 
Simon Clubley, clubley@remove_me.eisner.decus.org-Earth.UFP
Microsoft: Bringing you 1980s technology to a 21st century world

Re: Precisely why can't official FSF GNAT maintainers copy bug fixes in GNAT & its GCC-contained runtime en masse from GNAT GPL Community Edition?

[Dan'l Miller]

On Saturday, May 5, 2018 at 2:04:29 PM UTC-5, Dan'l Miller wrote:
> On Saturday, May 5, 2018 at 10:58:56 AM UTC-5, Lucretia wrote:
> > On Saturday, 5 May 2018 15:19:23 UTC+1, Dan'l Miller  wrote:
> > > On Saturday, May 5, 2018 at 7:44:54 AM UTC-5, Luke A. Guest wrote:
> > > >
> > > > GNAT Pro -> GNAT GPL -> FSF GNAT
> > > 
> > > Yes, that is how I suspect that it is as well.
> > > 
> > > The question is where precisely historically did each of the files is the GNAT-Pro branch originate?  I suspect that even the oldest of these are files whose 
> > 
> > Back in 1995, when I was learning Ada9X, GANT was distributed separately from GCC as it hadn't been accepted into the main tree, AFAIK. As of GCC-2.95.1 which is around EGCS era IIRC, there is no gcc/ada directory. I managed to find these:
> > 
> > http://archive.adaic.com/ase/ase02_02/lang/ada/gnat/3.12p/
> > http://sunsite.rwth-aachen.de:3080/ftp/pub/mirror/cs.nyu.edu/pub/gnat/3.15p/
> > 
> > You'll find there are copyright notices belonging to FSU as well there.
> > 
> > > rights to copy were assigned to FSF by NYU 1992 to 1993, likely predating even the formation of AdaCore (and its twin in France).  My suspicion is that even 
> > 
> > AdaCore was "spun out of NYU" and I still have no clue how it became a French company.
> 
> As shown in a 1996-era website
> http://www.cs.scranton.edu/~beidler/Ada/gnat/a-diocst.ads
> the assignment of copyright apparently started in 1992, at least for this file in the runtime.
> 
> AdaCore's own website states that AdaCore was launched in 1994 after the assignment of the rights to copy were assigned to FSF.
> 
> Release 1.83 of GNAT was prior to July 1994, directly from New York University.

On those old Walnut Creek Ada CDROMs (especially the March 1994 one), the 23 December 1993 release 1.67 of GNAT has a copyright notice that reads “Copyright (c) 1992, 1993, NYU”, apparently prior to the assignment to FSF (and definitely prior to the formation of AdaCore in 1994).  GNAT-the-compiler itself was licensed under the GPLv2.  (Perhaps the “modified” idea for the GMGPL was more of an idea from Florida State University than from NYU.)