Will the defect with formal package parameters be fixed in GNAT GPL 2009?

Will the defect with formal package parameters be fixed in GNAT GPL 2009?

[Eric Hughes]

Ten months ago in this newsgroup, I reported a defect in the handling
of formal package parameters in GNAT GPL 2008. The original posting is
available from Google Groups: http://groups.google.com/group/comp.lang.ada/browse_thread/thread/9aae240c681f9f62

My question, a very simple yes-or-no one, is this: Is this defect
corrected for a presumed 2009 release of GNAT GPL edition?

The defect precludes any particularly significant use of generic
programming in library code. The defect manifests itself when generic
packages are stacked up in a nested instantiation. Any serious use of
generic programming requires this approach. When I work in C++, it's
typical for me to nest template instantiations three or four levels
deep. I use generic programming for implementation polymorphism, where
I want the same visible semantics in different contexts (thus where
the total semantics must differ). A typical application is for a
second implementation that removes synchronization for a single-
tasking usage.

I ask because I'm soon planning on resurrecting the project I was
working on where I discovered the defect. If it's not fixed I'll be
switching to C++, and Ada will lose a design win.

Re: Will the defect with formal package parameters be fixed in GNAT GPL 2009?

[Vadim Godunko]

On Apr 27, 6:47 pm, Eric Hughes <eric....@gmail.com> wrote:
> Ten months ago in this newsgroup, I reported a defect in the handling
> of formal package parameters in GNAT GPL 2008. The original posting is
> available from Google Groups:http://groups.google.com/group/comp.lang.ada/browse_thread/thread/9aa...
>
> My question, a very simple yes-or-no one, is this: Is this defect
> corrected for a presumed 2009 release of GNAT GPL edition?
>
The answer depends from another question: did you report this bug to
AdaCore?

Re: Will the defect with formal package parameters be fixed in GNAT GPL 2009?

[Samuel Tardieu]

>>>>> "Vadim" == Vadim Godunko <vgodunko@gmail.com> writes:

Vadim> The answer depends from another question: did you report this bug
Vadim> to AdaCore?

Or to the FSF, on http://gcc.gnu.org/bugzilla if you also use the
pristine GNAT from GCC.

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

Re: Will the defect with formal package parameters be fixed in GNAT GPL 2009?

[Samuel Tardieu]

>>>>> "Sam" == Samuel Tardieu <sam@rfc1149.net> writes:

Sam> Or to the FSF, on http://gcc.gnu.org/bugzilla if you also use the
Sam> pristine GNAT from GCC.

Also note that the your example compiles fine using either GCC 4.3.3 or
later using

  with package S is new Signature (<>);

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

Re: Will the defect with formal package parameters be fixed in GNAT GPL 2009?

[Eric Hughes]

On Apr 27, 9:05 am, Vadim Godunko <vgodu...@gmail.com> wrote:
> The answer depends from another question: did you report this bug to
> AdaCore?

If you can show me where on the site http://libre.adacore.com/ there's
a place to report anything, I'd be pleased to see it. Insofar as I can
tell, AdaCore has no interest in engaging GPL users in any kind of
quality process. There's no way I could see to report defects, much
less any transparency about what might be done about such reports.
Actions speak louder than words, and in this case an absence of action
indicates an absence of interest.

Re: Will the defect with formal package parameters be fixed in GNAT GPL 2009?

[Eric Hughes]

On Apr 27, 9:31 am, Samuel Tardieu <s...@rfc1149.net> wrote:
> Or to the FSF, on http://gcc.gnu.org/bugzilla if you also use the
> pristine GNAT from GCC.

I don't use "pristine GNAT" because I use Windows, I have no patience
for system administration, and I use precompiled binaries. I get
precompiled binaries from the Libre site. I do not get precompiled
binaries from gnu.org.

Re: Will the defect with formal package parameters be fixed in GNAT GPL 2009?

[Eric Hughes]

On Apr 27, 9:36 am, Samuel Tardieu <s...@rfc1149.net> wrote:
> Also note that the your example compiles fine using either GCC 4.3.3 or
> later using
>
>   with package S is new Signature (<>);

It also compiles fine with GNU GPL 2008 and is beside the point. The
original example was cut down from existing code to a minimal example.
In the original, there were other named parameters along with
"( others => <> )", and so eliding the "other =>" part of the syntax
is not an option.

Re: Will the defect with formal package parameters be fixed in GNAT GPL 2009?

[Gautier]

Eric Hughes:

> If you can show me where on the site http://libre.adacore.com/ there's
> a place to report anything, I'd be pleased to see it. Insofar as I can
> tell, AdaCore has no interest in engaging GPL users in any kind of
> quality process. There's no way I could see to report defects, much
> less any transparency about what might be done about such reports.
> Actions speak louder than words, and in this case an absence of action
> indicates an absence of interest.

You may want to write to report@gnat.com . From my experience as a GPL 
user, they show always an excellent responsiveness - as long as you 
describe your issues very accurately, double-check your code examples. 
Unless your are absolutely sure that X is wrong, you'll better be 
diplomatic and write things like "I have the impression that X is wrong, 
as you can see with test code below". Well, as with every bug reporting, 
isn't it ?...
_________________________________________________________
Gautier's Ada programming -- http://sf.net/users/gdemont/
NB: For a direct answer, e-mail address on the Web site!

Re: Will the defect with formal package parameters be fixed in GNAT GPL 2009?

[Samuel Tardieu]

>>>>> "Eric" == Eric Hughes <eric.eh9@gmail.com> writes:

Eric> I don't use "pristine GNAT" because I use Windows, I have no
Eric> patience for system administration, and I use precompiled
Eric> binaries. I get precompiled binaries from the Libre site. I do not
Eric> get precompiled binaries from gnu.org.

There exist at least five ways to get GNAT:

 (1) Through a contract with AdaCore. This includes an very good
     commercial support.

 (2) From the libre.adacore.com site. This comes with no support unless
     you are an academic user qualifying for the GAP program.

 (3) From the Free Software Foundation, with a bug tracker and
     some volunteers working on bugs through an open process, provided
     that good bug reports are submitted.

 (4) From Cygwin, with precompiled Windows binaries, and the FSF GCC bug
     tracker can be used.

 (5) From various free software distribution vendors, such as Debian
     GNU/Linux, each with their own bug reporting system.

From these five options you chose the second one, which is the only one
with no support at all for non-academic users. Unless you are willing to
switch, you're on your own, and since you can't report your problem
through a proper channel, it is unlikely to get fixed unless someone in
another categorie reports it.

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

Re: Will the defect with formal package parameters be fixed in GNAT GPL 2009?

[Samuel Tardieu]

>>>>> "Eric" == Eric Hughes <eric.eh9@gmail.com> writes:

Eric> It also compiles fine with GNU GPL 2008 and is beside the
Eric> point. The original example was cut down from existing code to a
Eric> minimal example.  In the original, there were other named
Eric> parameters along with "( others => <> )", and so eliding the
Eric> "other =>" part of the syntax is not an option.

That may be a bug, but if you're not willing to accept workarounds for
the precise one you're describing, what is the point of asking for help?

You should have included one other named parameter so that it would have
been clear that the simple workaround I described cannot be applied to
your situation.

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

Re: Will the defect with formal package parameters be fixed in GNAT GPL 2009?

[sjw]

On Apr 27, 3:47 pm, Eric Hughes <eric....@gmail.com> wrote:

> My question, a very simple yes-or-no one, is this: Is this defect
> corrected for a presumed 2009 release of GNAT GPL edition?

When you posted this before, I asked what the Bugzilla (or other) ref
was that you'd posted the bug report under. You replied saying that
_I_ was welcome to post a bug report. I guess the gentle hint wasn't
enough

Newsflash: it's _your_ problem, _you_ post the bug report! (and if you
didn't, why are you surprised that it's not fixed?)

Re: Will the defect with formal package parameters be fixed in GNAT GPL 2009?

[Georg Bauhaus]

Eric Hughes wrote:
> On Apr 27, 9:05 am, Vadim Godunko <vgodu...@gmail.com> wrote:
>> The answer depends from another question: did you report this bug to
>> AdaCore?
> 
> If you can show me where on the site http://libre.adacore.com/ there's
> a place to report anything, I'd be pleased to see it. Insofar as I can
> tell, AdaCore has no interest in engaging GPL users in any kind of
> quality process.

When AdaCore staff made frequent appearences in public fora,
their interest in GNAT usage was quite visible. A stated
policy was to look at every bug report.

The channels might have changed, the web site and GPL edition doc
files might indeed be consistent with the No Support policy of
AdaCore's GNAT GPL edition and do not announce a reporting address.
However, the policy to look at every bug report that comes
in is, I guess, still true.
(Of the rare opportinities I had of reporting a bug report like
for a Bug Box (hopefully well dressed as hinted at by Gautier de
Montmollin in another posting) the majority got answers which
seemed to indicate that fixing was underway...)

I'd think that given the way the Lorenz code story is being
staged, they might indeed be interested in supporting a
promising project.

If memory serves, SofCheck once announced some interest
in open source uses of their error detection tool.
Don't know whether this could apply to their Ada frontend
as well, let alone whether or not there is capacity for
small scale support of alleged open source editions.

Re: Will the defect with formal package parameters be fixed in GNAT GPL 2009?

[Georg Bauhaus]

sjw wrote:
> On Apr 27, 3:47 pm, Eric Hughes <eric....@gmail.com> wrote:
> 
>> My question, a very simple yes-or-no one, is this: Is this defect
>> corrected for a presumed 2009 release of GNAT GPL edition?
> 
> When you posted this before, I asked what the Bugzilla (or other) ref
> was that you'd posted the bug report under. You replied saying that
> _I_ was welcome to post a bug report. I guess the gentle hint wasn't
> enough
> 
> Newsflash: it's _your_ problem, _you_ post the bug report! (and if you
> didn't, why are you surprised that it's not fixed?)

BTW, is it true that there a C++ compiler company that does
support open source projects with quickly fixed GPL compilers?

Re: Will the defect with formal package parameters be fixed in GNAT GPL 2009?

[Ludovic Brenta]

Samuel Tardieu wrote on comp.lang.ada:
> From these five options you chose the second one, which is the only one
> with no support at all for non-academic users. Unless you are willing to
> switch, you're on your own, and since you can't report your problem
> through a proper channel, it is unlikely to get fixed unless someone in
> another categorie reports it.

Note that you don't have to switch permanently; you only have to
reproduce the problem with one of the other four versions of GNAT and
report it in sufficient detail to the relevant bug tracking system.

--
Ludovic Brenta.

Re: Will the defect with formal package parameters be fixed in GNAT GPL 2009?

[Vadim Godunko]

On Apr 27, 8:59 pm, Eric Hughes <eric....@gmail.com> wrote:
> On Apr 27, 9:05 am, Vadim Godunko <vgodu...@gmail.com> wrote:
>
> > The answer depends from another question: did you report this bug to
> > AdaCore?
>
> If you can show me where on the sitehttp://libre.adacore.com/there's
> a place to report anything, I'd be pleased to see it.
$ gnatmake --help

<skip>

Report bugs to report@adacore.com

:-)

Re: Will the defect with formal package parameters be fixed in GNAT GPL 2009?

[Ludovic Brenta]

Vadim Godunko wrote on comp.lang.ada:
> Eric Hughes <eric....@gmail.com> wrote:
>> On Apr 27, 9:05 am, Vadim Godunko <vgodu...@gmail.com> wrote:
>>> The answer depends from another question: did you report this bug to
>>> AdaCore?
>
>> If you can show me where on the sitehttp://libre.adacore.com/there's
>> a place to report anything, I'd be pleased to see it.
>
> $ gnatmake --help
>
> <skip>
>
> Report bugs to rep...@adacore.com
>
> :-)

That's the canonical answer but, unfortunately, AdaCore's bug database
is private.  The other solutions that Samuel mentioned feature public
and searchable bug databases. I guess that, to get the best of both
worlds, the OP should report the problem both to AdaCore (for
resulution, hopefully, in the next GPL Edition) and to the GCC bug
database (to document the bug publicly and keep track of when the fix
lands into GCC).

--
Ludovic Brenta.

Re: Will the defect with formal package parameters be fixed in GNAT GPL 2009?

[Eric Hughes]

On Apr 27, 1:20 pm, sjw <simon.j.wri...@mac.com> wrote:
> Newsflash: it's _your_ problem, _you_ post the bug report! (and if you
> didn't, why are you surprised that it's not fixed?)

It's not my problem at all; I just won't be using Ada if it doesn't
work. C++ templates are perfectly adequate to satisfy what I need in
generic programming. I don't suppose that the Ada-only people here pay
much attention to C++, but there's some quite good work being done
there with generics, both in practice and in proposals for the next
language version. So there's exciting progress in C++ with generics,
and the only practical compiler for GPL code can't do a nested
instantiation. I really do fail to see how this is my personal
problem.

This is the Ada community's problem--not only the actual defect but
also the practical difficulty of even filing a report on it. It's a
problem because when you make things unnecessarily difficult, people
will go elsewhere. In the present case, my own position is quite clear
to me--I won't support a platform (= language + distributions +
organizations) by writing libraries for it when that platform throws
up barriers to entry and barriers to quality.

Re: Will the defect with formal package parameters be fixed in GNAT GPL 2009?

[Eric Hughes]

On Apr 27, 3:03 pm, Vadim Godunko <vgodu...@gmail.com> wrote:
> > $ gnatmake --help
>
> <skip>
>
> Report bugs to rep...@adacore.com

I'll grant that it's not completely and utterly absent, but I must
tell you that I'm very much in the habit of looking at the point of
distribution as also a point of reporting. I am very much not in the
habit in running random executables in the hopes of discovering such
information.

I am however, like you, amused at this.

Re: Will the defect with formal package parameters be fixed in GNAT GPL 2009?

[Eric Hughes]

On Apr 27, 12:23 pm, Samuel Tardieu <s...@rfc1149.net> wrote:
> There exist at least five ways to get GNAT:

Although I did say "precompiled binaries", which I admit might refer
only to the compiler, I use the distribution on Libre site because it
also provides an IDE. I first used GPS and most recently have used
GNATbench. Of the five ways of getting GNAT outlined, only the Libre
site provides IDE support also.

Re: Will the defect with formal package parameters be fixed in GNAT GPL 2009?

[Eric Hughes]

On Apr 27, 12:25 pm, Samuel Tardieu <s...@rfc1149.net> wrote:
> That may be a bug, but if you're not willing to accept workarounds for
> the precise one you're describing, what is the point of asking for help?

I wasn't asking for help. I was inquiring whether a defect was fixed
or not.

> You should have included one other named parameter so that it would have
> been clear that the simple workaround I described cannot be applied to
> your situation.

I suppose I could have done that, had I anticipated that people
wouldn't read that I had cut down the code to isolate the defect and
also anticipated that it wasn't already obvious that I knew of the
"workaround" by including it as a comment in the example provided.

Re: Will the defect with formal package parameters be fixed in GNAT GPL 2009?

[Adam Beneschan]

On Apr 28, 11:12 am, Eric Hughes <eric....@gmail.com> wrote:

> This is the Ada community's problem--not only the actual defect but
> also the practical difficulty of even filing a report on it.

I have some sympathy for your position.  If I encountered a bug, I'd
also look on the web site first to figure out how to get support,
report bugs, etc.  It wouldn't occur to me to try an -h or --help
option, as some other poster mentioned; the traditional meaning of
those options (in Unix-type systems) has been to tell me how to use a
program, not how to report bugs in it, so anyone who thinks you or I
"just should have known" to try that approach is wrong.

Nevertheless, you did post about the problem here ten months ago, and
you could have easily asked this same newsgroup how to report the bug
to the compiler maintainers, but you didn't.

                               -- Adam

Re: Will the defect with formal package parameters be fixed in GNAT GPL 2009?

[Björn Persson]

Eric Hughes wrote:

> This is the Ada community's problem

The Ada community's problem is a shortage of people who want to contribute.
There's no shortage of people who just want to whine. One way of
contributing is to report the bugs you find. Since you made the effort to
isolate the bug and produce a minimal test case, I don't understand why
you'd waste that effort instead of taking the last step and actually report
the bug.

It's generally assumed that the person who reports a bug is also the person
who can answer questions about the bug report. Reporting other people's
bugs is likely to cause unnecessary trouble. I once tried to be helpful and
entered a bug report in GCC's Bugzilla for a bug that someone else had
posted about here in comp.lang.ada. That report was closed unfixed because
of misunderstandings.

-- 
Bj�rn Persson
PGP key A88682FD

Re: Will the defect with formal package parameters be fixed in GNAT GPL 2009?

[Eric Hughes]

On Apr 28, 2:40 pm, Björn Persson <bj...@xn--rombobjrn-67a.se> wrote:
> The Ada community's problem is a shortage of people who want to contribute.
> There's no shortage of people who just want to whine.

As I see it, I'm the one with an expressed desire to contribute, and
you're the one whining about me. And not the only one, either. I'm
actually a little surprised about how many people think I'm at fault,
to blame, and/or deserve shame for not Doing-The-Right-Thing and
putting in Extra-Effort to report what seems to me to be a very basic
compiler defect.

Let me be clear. I am not now an Ada user. I have some interest in
becoming one again. All other things being equal, I would pick Ada
over C++. All other things are manifestly not equal. I have no
motivation to try to improve Ada as an outsider. My payback for such
activity is speculative at best. I am perfectly content, although a
bit disappointed, to see Ada decay toward oblivion.

AdaCore has a no support policy for GPL users, which means to me that
I expect to be ignored. Other with different experiences assure me
otherwise, yet these are not my experiences but theirs. I would be
perfectly happy to have someone else, perhaps one of those with
different experiences, lay claim to the defect and report it.

These days, not supporting the free software penumbra around a
platform is a recipe for slow failure. I've been in the position of
making platform selection and hiring a new programming team. There are
always a number of platforms that are more-or-less adequate and the
dominant factor in selecting one becomes availability of staff
competent in that platform. Without a free software environment for
people to self-train with, the number of available staff decreases.
For a language, like Ada, that's not ordinarily part of university
curriculum, such an environment can be the dominant mode of training.
Platforms are in _de facto_ competition for the attention of
programmers/potential-staff. The population dynamics for a platform
that treats its free software people with disdain is pretty clear--it
will decline.

Peter Drucker made the excellent point that if you want to improve
your sales, you should talk to your non-customers about why they don't
buy (rather than the ordinary practice of talking to your customers
about why they do). So here I am, a non-customer, as it were, of the
Ada platform. My advice is that listening is a more productive
response than blaming.

Re: Will the defect with formal package parameters be fixed in GNAT GPL 2009?

[Robert Matthews]

Eric Hughes wrote:


> These days, not supporting the free software penumbra around a
> platform is a recipe for slow failure. I've been in the position of
> making platform selection and hiring a new programming team. There are
> always a number of platforms that are more-or-less adequate and the
> dominant factor in selecting one becomes availability of staff
> competent in that platform. Without a free software environment for
> people to self-train with, the number of available staff decreases.
> For a language, like Ada, that's not ordinarily part of university
> curriculum, such an environment can be the dominant mode of training.
> Platforms are in _de facto_ competition for the attention of
> programmers/potential-staff. The population dynamics for a platform
> that treats its free software people with disdain is pretty clear--it
> will decline.
> 
> Peter Drucker made the excellent point that if you want to improve
> your sales, you should talk to your non-customers about why they don't
> buy (rather than the ordinary practice of talking to your customers
> about why they do). So here I am, a non-customer, as it were, of the
> Ada platform. My advice is that listening is a more productive
> response than blaming.

For us GPL-Ada people it seems that AdaCore is both our greatest
asset (via GNAT and related tools), and also our greatest problem.
I was greatly disappointed when I realized that AdaCore do not
coordinate their GPL releases with Ludovic. Also, I can remember
when Robert Dewar stated on cla that AdaCore generally ignore
bug reports from the GPL community.
For me, the ideal would be for AdaCore to coordinate their GPL
releases with the Ubuntu releases; I live in hope.

Robert

Re: Will the defect with formal package parameters be fixed in GNAT GPL 2009?

[Samuel Tardieu]

>>>>> "Robert" == Robert Matthews <ignored@ramatthews.free-online.co.uk> writes:

Robert> I was greatly disappointed when I realized that AdaCore do not
Robert> coordinate their GPL releases with Ludovic.

Ludovic doesn't use AdaCore's GPL releases at this time, he uses GNAT as
obtained by the FSF.

Robert> Also, I can remember when Robert Dewar stated on cla that
Robert> AdaCore generally ignore bug reports from the GPL community.

From my experience, they don't. They just classify them with a low
priority (compared to reports from paying customers). If a bug report
shows a real and crucial defect, you may be sure that they will fix it
in a short time so that their supported customers don't stumble upon it.

Robert> For me, the ideal would be for AdaCore to coordinate their GPL
Robert> releases with the Ubuntu releases; I live in hope.

Ubuntu has never been the best distribution for Ada. They get their
packages from Debian, which is today the top GNU/Linux distribution as
far as Ada development is concerned.

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

Re: Will the defect with formal package parameters be fixed in GNAT GPL 2009?

[Ludovic Brenta]

Eric Hughes wrote:
> AdaCore has a no support policy for GPL users, which means to me that
> I expect to be ignored. Other with different experiences assure me
> otherwise, yet these are not my experiences but theirs. I would be
> perfectly happy to have someone else, perhaps one of those with
> different experiences, lay claim to the defect and report it.

So, you expect AdaCore to ignore your bug reports, but expect people
on comp.lang.ada to report your bugs for you and not get ignored? I
fail to see the logic in this.

> These days, not supporting the free software penumbra around a
> platform is a recipe for slow failure. [...]

It seems to me as though you're not really complaining that AdaCore
does not support the "free software penumbra" (which you understand)
but that "the free software penumbra" does not support you.

AdaCore does support the free software "penumbra" by providing their
source code under terms of the GPL. The return on this investment
consists of new users and several external contributors who provide
additional software or improvements to the existing software. AdaCore
sell services such as packaging and bug fixes, so it would not be in
their best interests to provide the same service for free that they
provide to paying customers. You understand that and you expect to be
ignored but I can also understand that some here have taken offense at
your attitude consisting in (a) refusing to report your bug properly
in a bug database, (b) expecting other people to do that for you, and
(c) threatening not to use Ada if volunteers nobody fixes your bug.

> My advice is that listening is a more productive response than blaming.

My advice, if you want your bug fixed, is to get a support contract
from AdaCore. Failing that, you can still report your bug on GCC's
bugzilla database (http://gcc.gnu.org/bugzilla) and offer a bounty to
whoever fixes it for you.

--
Ludovic Brenta.

Re: Will the defect with formal package parameters be fixed in GNAT GPL 2009?

[Björn Persson]

Eric Hughes wrote:

> On Apr 28, 2:40�pm, Bj�rn Persson <bj...@xn--rombobjrn-67a.se> wrote:
>> The Ada community's problem is a shortage of people who want to
>> contribute. There's no shortage of people who just want to whine.
> 
> As I see it, I'm the one with an expressed desire to contribute, and
> you're the one whining about me.

Your post from June 2008 does look like an attempt to contribute by
reporting a bug. Unfortunately you chose the wrong channel for that. I'll
assume that was an honest mistake so I don't blame you for it. However,
your refusal to correct the mistake now that you know what the right
channels are indicates that your desire to contribute has gone away.

As for me, I contributed a binding to Libmilter less than two weeks ago.
Earlier I have contributed a command line parameter handler and an
encoding-aware string handling library. I have also reported several bugs
in GNAT. I'll let everyone judge for themselves on whether that makes me a
contributor or a whiner.

> I would be
> perfectly happy to have someone else, perhaps one of those with
> different experiences, lay claim to the defect and report it.

I won't make that mistake again, but if you enter a bug report at
http://gcc.gnu.org/bugzilla/ I will add a comment that I can reproduce it
with GCC 4.3.2 in Fedora 10.

-- 
Bj�rn Persson
PGP key A88682FD

Re: Will the defect with formal package parameters be fixed in GNAT GPL 2009?

[Ludovic Brenta]

Robert Matthews wrote on comp.lang.ada:
> I was greatly disappointed when I realized that AdaCore do not
> coordinate their GPL releases with Ludovic.

Thanks but I really am not the coordinating center of the Ada
universe. I follow the Debian and GCC release schedules; I don't make
them. Perhaps what you mean is that it would be nice if AdaCore would
follow the GCC release schedule; this I agree with but, of course, I
have no influence on AdaCore (and rightly so).

[...]

> For me, the ideal would be for AdaCore to coordinate their GPL
> releases with the Ubuntu releases; I live in hope.

There is too much software and too many distributions; coordinating
all releases with all distributions is impossible. That is why GNAT
GPL Edition is distribution-agnostic; you can install it on any
distribution.

--
Ludovic Brenta.

Re: Will the defect with formal package parameters be fixed in GNAT GPL 2009?

[Ludovic Brenta]

Samuel Tardieu wrote on comp.lang.ada:
> Ubuntu has never been the best distribution for Ada. They get their
> packages from Debian, which is today the top GNU/Linux distribution as
> far as Ada development is concerned.

Thanks a lot, Sam. Since this is back-patting time, I'll mention that
one of the many benefits of Debian is the backports of bug fixes into
Debian's default gnat compiler; Samuel is the author of most of these,
as is visible in both GCC's and Debian's bug databases. And Samuel was
the previous maintainer of Ada in Debian before Ubuntu even existed,
so he knows what he's talking about.

--
Ludovic Brenta.

Re: Will the defect with formal package parameters be fixed in GNAT GPL 2009?

[qunying]

Hi Eric,

I am not sure why you spend so much time arguing this which leads to
nowhere and yet you refuse to spend a few minutes to enter a bug
report?

Re: Will the defect with formal package parameters be fixed in GNAT GPL 2009?

[Eric Hughes]

On Apr 30, 10:14 am, qunying <zhu.quny...@gmail.com> wrote:
> I am not sure why you spend so much time arguing this which leads to
> nowhere and yet you refuse to spend a few minutes to enter a bug
> report?

Thank you for cutting to the chase.

I'm protesting. The AdaCore relationship to the free software world is
broken. I won't be using GNAT. It's not my problem, and I won't be
spending even a moment of my time fixing it myself.

I do see some worth, however speculative, in spurring people who care
to do something about it. Given the reactions I've received, that
worth is becoming increasingly speculative.

Re: Will the defect with formal package parameters be fixed in GNAT GPL 2009?

[Eric Hughes]

On Apr 30, 6:17 am, Ludovic Brenta <ludo...@ludovic-brenta.org> wrote:
> My advice, if you want your bug fixed, is to get a support contract
> from AdaCore.

The point is that don't really feel any need to have it fixed. I'm
happy to have it remained unfixed, in which case I'll do something
else.

Like many others who have replied, it seems to me that you're applying
the principle that "we're all in this together". If that were true,
efforts to modify my behavior would be perfectly appropriate. Yet I am
not in this together with all of you. I'm not wedded to Ada. I don't
have enough invested in it.

I'm an outsider at this point, delivering a message. I care only
enough to be a messenger. If you want to shoot the messenger, fine
with me, though after which you should anticipate fewer messages.

Re: Will the defect with formal package parameters be fixed in GNAT GPL 2009?

[Samuel Tardieu]

>>>>> "Eric" == Eric Hughes <eric.eh9@gmail.com> writes:

Eric> I'm protesting. The AdaCore relationship to the free software
Eric> world is broken. I won't be using GNAT. It's not my problem, and I
Eric> won't be spending even a moment of my time fixing it myself.

I get that, but why are you protesting *here*? None of the people
participating to this thread so far can do anything about AdaCore's
GNAT.

Eric> I do see some worth, however speculative, in spurring people who
Eric> care to do something about it. Given the reactions I've received,
Eric> that worth is becoming increasingly speculative.

I think many are puzzled by the energy you spend protesting here, while
you can probably get the bug fixed by reporting it through the proper
channel on the GCC site. This is where the community support takes
place, and code is synchronized on a regular basis between the FSF GCC
and AdaCore. A bug fixed in one of the code base will get fixed in the
other one.

You might not like this situation (AdaCore not providing free support
for the GPL version), but this is the way it is and most people are
satisfied with it. AdaCore has no obligation to provide free GNAT
releases, but they chose to do so as a community service[1]. You seem to
think that they should do more than that and provide free support as
well. I don't understand why.

  Sam

Notes: 
[1] And because it can only increase the number of Ada programmers,
    which is good for them.
-- 
Samuel Tardieu -- sam@rfc1149.net -- http://www.rfc1149.net/

Re: Will the defect with formal package parameters be fixed in GNAT GPL 2009?

[Robert A Duff]

Robert Matthews <ignored@ramatthews.free-online.co.uk> writes:

> ...Also, I can remember
> when Robert Dewar stated on cla that AdaCore generally ignore
> bug reports from the GPL community.

I think you are mis-remembering.  AdaCore accepts bug reports (via
report@adacore.com) from the general public.  If it turns out to
be a real bug, AdaCore will fix it eventually.

Paying customers of course have higher priority.

Perhaps you are thinking of the fact that AdaCore does not pay any
attention to comp.lang.ada.  Comp.lang.ada is just a discussion
forum -- if you want to report a bug, you have to submit it
through the right channels.  I read comp.lang.ada once in a while,
just for fun -- certainly not on a regular basis.  I think a few
other AdaCore employees do the same.  Most never read c.l.a.

- Bob

Re: Will the defect with formal package parameters be fixed in GNAT GPL 2009?

[Robert Matthews]

Robert A Duff wrote:

> I think you are mis-remembering.  AdaCore accepts bug reports (via
> report@adacore.com) from the general public.  If it turns out to
> be a real bug, AdaCore will fix it eventually.
> 

Sorry if my memory is not correct - this goes back some years now.

However, it is good to learn that report@adacore.com is a useful
channel, though it would be nice to see this on the libre web site
https://libre.adacore.com/

Robert.

Re: Will the defect with formal package parameters be fixed in GNAT GPL 2009?

[Robert Matthews]

Robert A Duff wrote:

> Perhaps you are thinking of the fact that AdaCore does not pay any
> attention to comp.lang.ada.  Comp.lang.ada is just a discussion
> forum -- if you want to report a bug, you have to submit it
> through the right channels.  I read comp.lang.ada once in a while,
> just for fun -- certainly not on a regular basis.  I think a few
> other AdaCore employees do the same.  Most never read c.l.a.
> 

I note that cla seems to be the only forum for discussing
matters of mutual interest, therefore AdaCore's lack of interest in
cla does seem to be a shame.

Robert.

Re: Will the defect with formal package parameters be fixed in GNAT GPL 2009?

[Georg Bauhaus]

Eric Hughes schrieb:
> On Apr 30, 10:14 am, qunying <zhu.quny...@gmail.com> wrote:
>> I am not sure why you spend so much time arguing this which leads to
>> nowhere and yet you refuse to spend a few minutes to enter a bug
>> report?
> 
> Thank you for cutting to the chase.
> 
> I'm protesting. The AdaCore relationship to the free software world is
> broken. I won't be using GNAT. It's not my problem, and I won't be
> spending even a moment of my time fixing it myself.
> 
> I do see some worth, however speculative, in spurring people who care
> to do something about it. Given the reactions I've received, that
> worth is becoming increasingly speculative.

The speculation now is that a flaw or a bug in a compiler
can indeed become a blinking spot on the makers' radar screen,
even when posted to comp.lang.ada.

E.g.,

http://libre.adacore.com/2009/05/15/NF-63-I511-028-gnat/

(About System.OS_Lib when GNAT.OS_Lib should be used,
as discussed here recently.)