gnat-3.10

gnat-3.10

[Ronald Cole]

Robert,

Are we going to see a 3.10 release which can bootstrap itself on HPUX
(3.09 can't) anytime soon?

-- 
Forte International, P.O. Box 1412, Ridgecrest, CA  93556-1412
Ronald Cole <ronald@ridgecrest.ca.us>    Phone: (760) 499-9142
President, CEO                             Fax: (760) 499-9152
My PGP fingerprint: E9 A8 E3 68 61 88 EF 43  56 2B CE 3E E9 8F 3F 2B

Re: gnat-3.10

[Robert Dewar]

Ronald Cole

<<Robert,

Are we going to see a 3.10 release which can bootstrap itself on HPUX
(3.09 can't) anytime soon?>>


If you see a public release of GNAT from us, then you know it has been
bootstrapped, i.e. we succeeded in bootstrapping the HPUX 3.09 version.
It's not too surprising that Ronald might not be able to do so without
help from us, there can be tricky points involved in any of these GNAT
bootstraps, which is why most people work with the binary releases.

As for 3.10, we never give any prognosis on schedules for public releases.
A number of our customers are working with prereleases of 3.10 very
successfully, and all we will say is that, as for all GNAT releases, it
will be releases publicly some time in the future.

Of course we cannot guarantee that when it is realeased, Ronald Cole
will be able to bootstrap it on HPUX!

(but we will have bootstrapped it!)

Re: gnat-3.10

[Ronald Cole]

dewar@merv.cs.nyu.edu (Robert Dewar) writes:
> If you see a public release of GNAT from us, then you know it has been
> bootstrapped, i.e. we succeeded in bootstrapping the HPUX 3.09 version.
> It's not too surprising that Ronald might not be able to do so without
> help from us, there can be tricky points involved in any of these GNAT
> bootstraps, which is why most people work with the binary releases.

Obviously you personally haven't tried to build gnat-3.09-src.tar.gz
with gnat-3.09-hppa1.1-hp-hpux10.10-bin.tar.gz while following the
directions exactly as given in src/README.UNIX.  Please!  Try it for
yourself and be sure to post your results.  Even with the following
patch provided by your team, the gnat1 component bootstrapped itself,
but it couldn't finish building the Run Time Library:

--- function.c~ Tue Feb 11 13:33:53 1997
+++ function.c  Tue Feb 18 12:32:05 1997
@@ -2963,5 +2963,5 @@
                                          0))
          || (GET_CODE (XEXP (x, 0)) == REG
-             && instantiate_virtual_regs_1 (&XEXP (x, 0), 0, 0)))
+             && instantiate_virtual_regs_1 (&XEXP (x, 0), object, 0)))
        return 1;


> As for 3.10, we never give any prognosis on schedules for public releases.
> A number of our customers are working with prereleases of 3.10 very
> successfully, and all we will say is that, as for all GNAT releases, it
> will be releases publicly some time in the future.
>
> Of course we cannot guarantee that when it is realeased, Ronald Cole
> will be able to bootstrap it on HPUX!
> (but we will have bootstrapped it!)

I hope you'll understand that I'll still question the validity of a
binary release that can't even compile itself without asserting a
fatal compiler error.

-- 
Forte International, P.O. Box 1412, Ridgecrest, CA  93556-1412
Ronald Cole <ronald@ridgecrest.ca.us>    Phone: (760) 499-9142
President, CEO                             Fax: (760) 499-9152
My PGP fingerprint: E9 A8 E3 68 61 88 EF 43  56 2B CE 3E E9 8F 3F 2B

Re: gnat-3.10

[Robert Dewar]

Ronald Cole says

<<I hope you'll understand that I'll still question the validity of a
binary release that can't even compile itself without asserting a
fatal compiler error.
>>

It is perfectly possible to bootstrap this version of GNAT, if you have
the right versions of everything. So the issue here is not whether this
version can compile itself, of course it can, it is whether Ronald Cole
without help from us can do it? Apparently the answer is no. We provide
our binary releases on an as-is basis, and they have always been bootstrapped.
We do NOT particularly spend a lot of effort ensuring that they can easily
be built without our help from sources, though many people do in fact succeed
in this effort.

Certainly if you are specifically concerned with the "validity" of a binary
release, we would advise you to get it from us. We take no responsibility
for public versions that are around on the net. We try to make sure that
they are in good shape, and correspond to what we distributed originally, but
we cannot guarantee this, and we certainly do NOT guarantee that everyone will
be able to build them from sources.

Ada Core Technologies
Robert B. K. Dewar

Re: gnat-3.10

[Robert Dewar]

Ronald Cole said

<<Bootstrapped according to the directions you so thoughtfully provide
in your source distribution?  Doubtful, see below.
 >>

Well some people manage with these directions alone, by fiddling around,
other people succeed with advice that we give. We do not provide any help
in building from sources for unsupported users of GNAT. Nevertheless, many
such people succeed in building from sources. And, as I have said before,
any binary version you see from us has been built from sources and
bootstrapped.

Re: gnat-3.10

[Stephen Leake]

Ronald Cole, in a fit of frustration, posted a long script documenting
his attempt to install and bootstrap gnat.

I am one of the "other people" that have successfully retargeted GNAT,
without bugging ACT for free help.

Ronald; this does not belong in this news group. If you are a paying
customer of ACT, use email with them. If you are not a paying customer,
give up! They are generously making the binaries available, with minimal
instructions on how to change them around. If you cannot do what you
want with the material provided, then pay the support cost, and get the
help you need. ACT is not in business to lose money, and the rest of us
would like them to keep providing free binaries!
-- 
- Stephe

Re: gnat-3.10

[Ronald Cole]

dewar@merv.cs.nyu.edu (Robert Dewar) writes:
> Ronald Cole said
> <<Bootstrapped according to the directions you so thoughtfully provide
> in your source distribution?  Doubtful, see below.
>  >>
> 
> Well some people manage with these directions alone, by fiddling around,
> other people succeed with advice that we give. We do not provide any help
> in building from sources for unsupported users of GNAT. Nevertheless, many
> such people succeed in building from sources. And, as I have said before,
> any binary version you see from us has been built from sources and
> bootstrapped.

Why do your excuses keep changing, Bob?  And why didn't you answer my
question at the end of my post?  I think we all know why...

I've clearly demonstrated that the gnat1 binary in the hpux
distribution is broken... for both your supported *and* unsupported
users.  Plain and simple.  Of course, you're too proud to fix it or to
document the actual "fiddling around" required to bootstrap on this
platform in the src/README.UNIX file because that would be tantamount
to admitting that you might actually have been wrong.

-- 
Forte International, P.O. Box 1412, Ridgecrest, CA  93556-1412
Ronald Cole <ronald@ridgenet.net>        Phone: (760) 499-9142
President, CEO                             Fax: (760) 499-9152
My PGP fingerprint: E9 A8 E3 68 61 88 EF 43  56 2B CE 3E E9 8F 3F 2B

Re: gnat-3.10

[Robert Dewar]

Ronald Cole said

<<Why do your excuses keep changing, Bob?  And why didn't you answer my
question at the end of my post?  I think we all know why...
 
I've clearly demonstrated that the gnat1 binary in the hpux
distribution is broken... for both your supported *and* unsupported
users.  Plain and simple.  Of course, you're too proud to fix it or to
document the actual "fiddling around" required to bootstrap on this
platform in the src/README.UNIX file because that would be tantamount
to admitting that you might actually have been wrong.>>

First, I am Robert, not Bob :-)

Second, we do our best to give accurate directions for building from
sources. Generally we are quite successful, since many knowledgable
people can indeed build successfully, even if they are not our customers.
Of course if they are our customers, then we provide whatever help they
need.

Your difficulties might be incompetence, or something you have overlooked,
or a glitch in the instructions, or a wrong patch somewhere, or a wrong
version of something, but the point is that we are not the slightest bit
interested in spending time to investigate which on a volunteer basis.

I do not know quite what you want me to admit as wrong. All I have every
claimed is that a lot of people have managed to build from sources, and that
we do not care to spend any time helping people build from sources unless
they are our customers, and finally, that all binary versions of GNAT have
been bootstrapped by us (it is actually impossible for us to prepare a binary
version without bootstrapping, since the process of preparing a binary version
is part of the run that does a bootstrap).

The fact that you cannot build from sources would be a concern to us if you
were a customer, but since you are not, it is not. As i said before, we do
our best to make the instructions for building from sources complete and
accurate, but we make absolutely NO guarantees that these instructions are
complete or accurate.

You seen to continue to expect me or us to spend time looking through your
build attempts. Sorry, won't happen if you are not a customer! We are a
commercial company, not a help-Ronald-with-his-problems-at-no-cost
organization!

Robert Dewar
Ada Core Technologies

Re: gnat-3.10

[Robert Dewar]

Ronald Cole said

<<<<Why do your excuses keep changing, Bob?  And why didn't you answer my
question at the end of my post?  I think we all know why...>>

Just to be clear, I do indeed hope that we all know why, the reason is that
I did not even read through your post, and I do not intend to (and I can't
understand why you expect me to!)

Robert Dewar
Ada Core Technologies

Re: gnat-3.10

[Jerry van Dijk]

In article <m2vi3kpuay.fsf@devo.ridgenet.net> ronald@ridgenet.net writes:

>$ uname -a
>HP-UX pitbull B.10.10 A 9000/735
>$ ftp cs.nyu.edu
>Connected to cs.nyu.edu.

etc, etc,

Couldn't this be sent through email ? Some of us have to pay for reading
c.l.a.

Jerry.

--

-- Jerry van Dijk  | Leiden, Holland
-- Consultant      | Team Ada
-- Ordina Finance  | jdijk@acm.org

Re: gnat-3.10

[Ronald Cole]

dewar@merv.cs.nyu.edu (Robert Dewar) writes:
> First, I am Robert, not Bob :-)

Sorry, you are correct.  You have no slack.

> Second, we do our best to give accurate directions for building from
> sources. Generally we are quite successful, since many knowledgable
> people can indeed build successfully, even if they are not our customers.
> Of course if they are our customers, then we provide whatever help they
> need.

That was your best?  Ok.  I'll take your word for it.

> Your difficulties might be incompetence, or something you have overlooked,
> or a glitch in the instructions, or a wrong patch somewhere, or a wrong
> version of something, but the point is that we are not the slightest bit
> interested in spending time to investigate which on a volunteer basis.

Might be?  I could have sworn you said it was without doubt.  Please
check your ego at the door, Robert.

> I do not know quite what you want me to admit as wrong. All I have every
> claimed is that a lot of people have managed to build from sources, and that
> we do not care to spend any time helping people build from sources unless
> they are our customers, and finally, that all binary versions of GNAT have
> been bootstrapped by us (it is actually impossible for us to prepare a binary
> version without bootstrapping, since the process of preparing a binary version
> is part of the run that does a bootstrap).

I have successfully built from sources up to 3.07.  After 3.07, some
patches were introduced that apparently broke -O optimization for the
pa-risc platform (-O2 optimization has been broken, and documented as
such, since at least 2.04).  I should probably tell you that compiling
from sources with -g only won't bootstrap either.  In 3.09, the binary
release end-user is forced to examine the object code to see whether
one gets correctly compiled code with -O or with -g.  You are
apparently very proud of this result.

> The fact that you cannot build from sources would be a concern to us if you
> were a customer, but since you are not, it is not. As i said before, we do
> our best to make the instructions for building from sources complete and
> accurate, but we make absolutely NO guarantees that these instructions are
> complete or accurate.

I see.  Apparently, you have no customers on the hpux platform as of
the 3.09 binary release.

> You seen to continue to expect me or us to spend time looking through your
> build attempts. Sorry, won't happen if you are not a customer! We are a
> commercial company, not a help-Ronald-with-his-problems-at-no-cost
> organization!

My post was a bug report, and a suggestion that perhaps you not
distribute binaries with broken optimizers for the 3.10 release as you
did with the 3.09 release.  Apparently (being money-grubbers), ACT
cannot assure quality, but only ensure the lack thereof in some of
it's binary releases.  I accept your plea of ignorance in this regard.

-- 
Forte International, P.O. Box 1412, Ridgecrest, CA  93556-1412
Ronald Cole <ronald@ridgenet.net>        Phone: (760) 499-9142
President, CEO                             Fax: (760) 499-9152
My PGP fingerprint: E9 A8 E3 68 61 88 EF 43  56 2B CE 3E E9 8F 3F 2B

Re: gnat-3.10

[Ronald Cole]

dewar@merv.cs.nyu.edu (Robert Dewar) writes:
> Just to be clear, I do indeed hope that we all know why, the reason is that
> I did not even read through your post, and I do not intend to (and I can't
> understand why you expect me to!)

Robert, you posted your fallacious premises and I proved them wrong in
considerable detail.

You alone chose to wallow in your own abject ignorance because you're
too prejudiced to accept bug reports from people you bully in this
forum, too proud to admit that you made a sub-standard binary
release, and too unconcerned about repeating that mistake.

If all you have to offer in rebuttal is more ad-hominem attacks, please
don't bother: it's not very professional of you.

-- 
Forte International, P.O. Box 1412, Ridgecrest, CA  93556-1412
Ronald Cole <ronald@ridgenet.net>        Phone: (760) 499-9142
President, CEO                             Fax: (760) 499-9152
My PGP fingerprint: E9 A8 E3 68 61 88 EF 43  56 2B CE 3E E9 8F 3F 2B

Re: gnat-3.10

[Dale Pontius]

In article <m2d8pms722.fsf@devo.ridgenet.net>,
        Ronald Cole <ronald@ridgenet.net> writes:
>
> too prejudiced to accept bug reports from people you bully in this

You seem to mistake postings here for official bug reports. Read
ANY documentation on GNAT, and you'll find out that it is provided
as-is. If support is required, and if GNAT is a critical part of
your business, then you should become a paying GNAT user. Then you
will get support.

More importantly, ACT and Cygnus, among others are an absolutely
critical experiment in the software industry. They will answer the
question, "Is there any middleground between students/hobbyists
hacking away at GNU code and commercial software houses like Micro-
soft?" Only a little bit of GNU code has made it to the mainstream,
and it's still pretty well untrusted in the business world, and
it's all because of "support." Even though informal Usenet support
from the author/maintainer is usually better than any company,
business demands a business. ACT/Cygnus/etc are an important bridge
that retains free software, yet makes it acceptable to business.

You call ACT money-grubbing because they're not giving you free
support. I call them downright generous because I have access to
their software for free, even if it is unsupported.

Dale Pontius
(NOT speaking for IBM)

Re: gnat-3.10

[CHARLET Arnaud]

Ronald Cole (ronald@ridgenet.net) wrote:
: Robert, you posted your fallacious premises and I proved them wrong in
: considerable detail.

You only proved you weren't able to compile gnat, that's all.
How long will you post your uninteresting and angry messages on this
newsgroup ? Apparently, you don't know what is compilation/gcc/gnat/
Makefiles, maybe you should buy some books on it and read them.

: You alone chose to wallow in your own abject ignorance because you're
: too prejudiced to accept bug reports from people you bully in this
: forum, too proud to admit that you made a sub-standard binary
: release, and too unconcerned about repeating that mistake.

Please stop being angry and stop posting your long and uninteresting
messages, lots of people pay for reading news.

: If all you have to offer in rebuttal is more ad-hominem attacks, please
: don't bother: it's not very professional of you.

Ok kid, I don't want to know if/where you work, but I'm glad I'm not
your boss.

Arno

Re: gnat-3.10

[Joerg Rodemann]

Hello folks!

Ronald Cole (ronald@ridgenet.net) wrote:
> Why do your excuses keep changing, Bob?  And why didn't you answer my
> question at the end of my post?  I think we all know why...

> I've clearly demonstrated that the gnat1 binary in the hpux
> distribution is broken... for both your supported *and* unsupported
> users.  Plain and simple.  Of course, you're too proud to fix it or to
> document the actual "fiddling around" required to bootstrap on this
> platform in the src/README.UNIX file because that would be tantamount
> to admitting that you might actually have been wrong.

Makes me thinking...wouldn't it be nice if EVERYTHING I am unable to install,
to build or to develop --- is definitely broken? For if I can't do it, nobody
ca do it? 

I think rather no. Maybe building GNAT is not an easy thing, I do not know. 
Although I managed a whole lot of software installations also from
source until now I failed to build GNAT. Most problably this was due to
lack of both time and experience (that was also my first try to compile gcc).

On the otherhand a collegue of mine sure succeded in compiling, linking and
installing gcc with C/C++ and Ada-support (That was GNAT 3.03 or 3.04.) within
two days. And it only took him that long because the machine was not the 
fastest at that time. As far as I remember he too had some problems which
occured due to usage of the wrong optimization level. Sure it was explained
in the installation instructions. :-} We just did not look at it carefully
enough.

So is GNAT 3.10 broken? I very strongly believe this is not the case: using 
a lot of binary versions for quite a time now --- on different plattform ---
I am sure the binaries are working. And those binaries must be built from
some source, don't they? And I can't imagine that Mr. Dewar or anyone else at 
ACT will change their releases just in order to leave YOU unable to build that
release. Especially if there are some folks who tell you that they've 
succeeded. 

Well, just a few thoughts...if I am able to admit something is or at least
was beyond my capabilies --- why can't you? Just blaming other people does
not help you very much, especially if you do it in such an aggressive and 
unpolite way. 

Have nice day

Yours

Joerg

--
Dipl.-Phys. Joerg S. Rodemann                  rodemann@mathematik.uni-ulm.de
Flurstrasse 21      ---     D-70372 Stuttgart     ---     Tel. (0711) 5090670

       !!!    My former address at Blaustein is no longer valid     !!!

Re: gnat-3.10

[Robert Dewar]

Dale says

<<More importantly, ACT and Cygnus, among others are an absolutely
critical experiment in the software industry. They will answer the
question, "Is there any middleground between students/hobbyists
hacking away at GNU code and commercial software houses like Micro-
soft?" Only a little bit of GNU code has made it to the mainstream,
and it's still pretty well untrusted in the business world, and
it's all because of "support." Even though informal Usenet support
from the author/maintainer is usually better than any company,
business demands a business>>

The reason that business demands a business is that they need to be sure
they have support when they need it, and that they get to decide when the
need for support arises. It is interesting that more and more of our support
is not related to bugs in GNAT, but rather, as people build more complex
applications using GNAT, and using the more complex features of Ada 95,
with giving people assistance in the use of GNAT. Also a lot of our
business relates to special purpose needs.

Yes, sometimes informal Usenet support can be quite effective, but on the
other hand, it can sometimes be highly unreliable. In particular, people
giving free support can quite reasonably take the attitude that you should
spend a lot of your time creating an absolutely clear example, which can be
a very time consuing excercise.

We quite often deal with large programs which don't seem to work at some 
particular point, and we dig around. Sometimes the digging shows up a GNAT
bug, sometimes a bug in the customer code, sometimes just a misunderstanding.
This digging can of course be very time consuming.

Also, I think the informal model works best when you have a program that is
of a level that one person can maintain it as a spare time hobby. Free software
has faired quite well, even without formal support in this kind of situation.

However, this model does not work well for large complex pieces of software,
like GCC, or GNAT, where you need a team of people working full time. In such
cases, the Cygnus or ACT model seems more workable (a similar situation is
true with Linux/GNU where more than one company has sprung up to support
this system.

Re: gnat-3.10

[Ronald Cole]

charlet@archimede.enst-bretagne.fr (CHARLET Arnaud) writes:
> You only proved you weren't able to compile gnat, that's all.
> How long will you post your uninteresting and angry messages on this
> newsgroup ? Apparently, you don't know what is compilation/gcc/gnat/
> Makefiles, maybe you should buy some books on it and read them.

Perhaps it is you that should take your own advice.  I proved that
the hpux gnat1 3.09 binary cannot compile sem-case.adb with -O
optimization as the directions say to compile it for HPUX.  I did
this because Robert questioned my ability to follow his directions.

> Ok kid, I don't want to know if/where you work, but I'm glad I'm not
> your boss.

Please brush up on your English.  This sentence of yours makes no sense.

-- 
Forte International, P.O. Box 1412, Ridgecrest, CA  93556-1412
Ronald Cole <ronald@ridgenet.net>        Phone: (760) 499-9142
President, CEO                             Fax: (760) 499-9152
My PGP fingerprint: E9 A8 E3 68 61 88 EF 43  56 2B CE 3E E9 8F 3F 2B

Re: gnat-3.10

[Ronald Cole]

pontius@btv.vnet.ibm.com (Dale Pontius) writes:
> You seem to mistake postings here for official bug reports. Read
> ANY documentation on GNAT, and you'll find out that it is provided
> as-is. If support is required, and if GNAT is a critical part of
> your business, then you should become a paying GNAT user. Then you
> will get support.

Please learn to follow a thread.  I posted early on that I reported
this bug several months ago and got back a patch from Richard Kenner
but that it didn't seem to do the trick.

> You call ACT money-grubbing because they're not giving you free
> support. I call them downright generous because I have access to
> their software for free, even if it is unsupported.

I call them money-grubbing because Robert is constantly bringing it
up.  "I'll tell you it's broken it if you pay me" probably isn't such
a hot marketing strategy...  Nowhere did I say I needed it fixed now;
but, it will need to be fixed if ACT is ever to have customers on the
pa-risc platform.

In fact, it may already be fixed; but I'm guessing it isn't.  My
original post only attempted to asked if ACT will have time to address
it for the 3.10 release, because they clearly didn't for the 3.09
release.

Robert then replied that there was nothing wrong with the 3.09 release
and proceeded to spew venom at me; forcing me to demonstrate that, yes
indeed, there is something wrong with the 3.09 release.

-- 
Forte International, P.O. Box 1412, Ridgecrest, CA  93556-1412
Ronald Cole <ronald@ridgenet.net>        Phone: (760) 499-9142
President, CEO                             Fax: (760) 499-9152
My PGP fingerprint: E9 A8 E3 68 61 88 EF 43  56 2B CE 3E E9 8F 3F 2B

Re: gnat-3.10

[Ronald Cole]

rodemann@mathematik.uni-ulm.de (Joerg Rodemann) writes:
> Makes me thinking...wouldn't it be nice if EVERYTHING I am unable to install,
> to build or to develop --- is definitely broken? For if I can't do it, nobody
> ca do it? 

I never said I was unable to install their release.  I only said that
the gcc-272.dif file seems to contain patches that break the -O level
optimization for some files.  (If you throw in the patches that come
with g77-0.5.20, the stage2 gnat1 coughs up sig 11 on every
invocation.)  This is not insignificant since it was the only
optimization level that could be used throughout to build GNAT on the
pa-risc for all 3.xx releases up to and including 3.07.

-- 
Forte International, P.O. Box 1412, Ridgecrest, CA  93556-1412
Ronald Cole <ronald@ridgenet.net>        Phone: (760) 499-9142
President, CEO                             Fax: (760) 499-9152
My PGP fingerprint: E9 A8 E3 68 61 88 EF 43  56 2B CE 3E E9 8F 3F 2B

Re: gnat-3.10

[Pascal Obry]

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1: Type: text/plain, Size: 1287 bytes --]



Well speaking for myself, I have been able to compile (and bootstrap) GNAT
since version 1.x. This is not hard at all if you follow the instructions.
I've been
using -O2 to build GNAT itself and I've never experienced any problem...

I've done this under Solaris and SUN-OS so maybe HPUX is not as easy but
the instructions given by ACT was correct, so why should they gave us
wrong
instructions for HPUX ? Anyway they provide a binary distribution for HPUX
so
they were able to build it no !

Pascal.
--

--|------------------------------------------------------------
--| Pascal Obry                               Team-Ada Member |
--|                                                           |
--| EDF-DER-IPN-SID- Ing�nierie des Syst�mes d'Informations   |
--|                                                           |
--| Bureau G1-010           e-mail: pascal.obry@der.edfgdf.fr |
--| 1 Av G�n�ral de Gaulle  voice : +33-1-47.65.50.91         |
--| 92141 Clamart CEDEX     fax   : +33-1-47.65.50.07         |
--| FRANCE                                                    |
--|------------------------------------------------------------
--|
--|   http://ourworld.compuserve.com/homepages/pascal_obry
--|
--|   "The best way to travel is by means of imagination"

Re: gnat-3.10

[Robert Dewar]

Ronald Cole saysw

<<I see.  Apparently, you have no customers on the hpux platform as of
the 3.09 binary release.
 
> You seen to continue to expect me or us to spend time looking through your
> build attempts. Sorry, won't happen if you are not a customer! We are a
> commercial company, not a help-Ronald-with-his-problems-at-no-cost
> organization!
 
My post was a bug report, and a suggestion that perhaps you not
distribute binaries with broken optimizers for the 3.10 release as you
did with the 3.09 release.  Apparently (being money-grubbers), ACT
cannot assure quality, but only ensure the lack thereof in some of
it's binary releases.  I accept your plea of ignorance in this regard.
>>


We have a number of large customers on HPUX. They were using 3.09 when it
was the current release, most of them have by now switched to 3.10 which is
the current product release. As usual, this will be released publicly in
the future. I must say that the logic that says "I, Ronald Cole, have failed
in attempting to build from sources, therefore no other customers are using
this successfully", entirely elludes me.

If your definition of money-grubber is someone who refuses to provide you
free help and assistance, yes, I am afraid that we fall into this category.
I imagine your local supermarket also is guilty.

As for ignorance. Yes, I am ignorant of your specific problem, because we
have not looked at it. I simply deleted your long post. If indeed you think
of it as a bug report, then bug reports get submitted to report@gnat.com.
Posting them to CLA will not help get them fixed!

Robert Dewar
Ada Core Technologies

Re: gnat-3.10

[Ronald Cole]

Andrew Lynch <lynch@cci.de> writes:
> I see being able to get a free version of GNAT as a privilege, not a
> right. If GNAT does not work for me, then tough shit.

I am under the impression GNAT is under the GPL because it was
developed with government money and uses the gcc backend.  The "tough
shit" part really isn't true because source is made available.

> I can either have a look at the sources and try to fix it myself, or
> I can *pay* ACT to make me a version that works for me, or I can
> wait for the changes that somebody else *paid* ACT to make become
> available in a public release.

The latter is what I'm doing since Robert has convinced me that I
probably should charge him for any fixes I make to his product.  And
since it'll be a cold day in hell when that happens, I won't be able
to distribute fixed versions.  It looks like quite a bit of work to
make the pa-risc more robust with Ada.  It also seems like no one is
willing to pay the high cost to fix it.  Of course, I'm not screwing
around with any of the gcc-2.8 pre-releases so I may be completely
off-base here, but no one has provided any information to indicate
that it will change in this area.

I would be very willing to pay for support once I saw that the product
had matured on that platform.  Robert has yet to convince me with a
factual argument that GNAT for the pa-risc has.

> For my personal use of GNAT I usually go with the third option. If
> you want to use GNAT in your company I would suggest you go with
> option two and carry on these arguments via email.

No client of mine would be willing to pay what it would cost to make
GNAT more robust on the pa-risc platform.

> Maybe ACT could fix every problem free of charge, but for how long?
> At some point all ACT employees will have starved to death and then
> where will you send your bug reports..?

The Hacker Ethic that spawned the GNU Manifesto has both feet in the
coffin.  I contributed a considerable amount and time and effort to
develop COFF support for and get it into g++-1.32 almost a decade ago.
It was the right thing to do and I don't regret not profiting from it
at all.

-- 
Forte International, P.O. Box 1412, Ridgecrest, CA  93556-1412
Ronald Cole <ronald@ridgenet.net>        Phone: (760) 499-9142
President, CEO                             Fax: (760) 499-9152
My PGP fingerprint: E9 A8 E3 68 61 88 EF 43  56 2B CE 3E E9 8F 3F 2B

Re: gnat-3.10

[Samuel Tardieu]

Arnaud> Ok kid, I don't want to know if/where you work, but I'm glad
Arnaud> I'm not your boss.

Ronald> Please brush up on your English.  This sentence of yours makes
Ronald> no sense.

When you have nothing more to say about the problem and do not want to 
stop then start criticizing the style...

  Sam
-- 
Samuel Tardieu -- sam@ada.eu.org

Re: gnat-3.10

[Andrew Lynch]

Ronald Cole wrote:
> pontius@btv.vnet.ibm.com (Dale Pontius) writes:
[...]
> > You call ACT money-grubbing because they're not giving you free
> > support. I call them downright generous because I have access to
> > their software for free, even if it is unsupported.
> 
> I call them money-grubbing because Robert is constantly bringing it
> up.  "I'll tell you it's broken it if you pay me" probably isn't such
> a hot marketing strategy...

Indeed it is not. A much better marketing strategy would be:
1) "I'll give you the product if you pay me"
2) "I'll tell you it's broken if you pay me some more"
3) "I'll fix what is broken if you pay me even more"
4) "I'll actually let you have the fixed version if you pay me..."

Many software and hardware companies run successfully like this.
I've been told that some places (maybe MS or Compaq?) even ask you
to pay for _sending_ them bug reports, let alone somebody actually
dealing with the bug.

I see being able to get a free version of GNAT as a privilege, not a
right. If GNAT does not work for me, then tough shit. I can either
have a look at the sources and try to fix it myself, or I can *pay*
ACT to make me a version that works for me, or I can wait for the
changes that somebody else *paid* ACT to make become available in a
public release.
For my personal use of GNAT I usually go with the third option. If
you want to use GNAT in your company I would suggest you go with
option two and carry on these arguments via email.

Andrew.
(Beware of sarcasm)

Maybe ACT could fix every problem free of charge, but for how long?
At some point all ACT employees will have starved to death and then
where will you send your bug reports..?

Re: gnat-3.10

[Richard Kenner]

In article <dewar.866559209@merv> dewar@merv.cs.nyu.edu (Robert Dewar) writes:
>Yes, sometimes informal Usenet support can be quite effective, but on the
>other hand, it can sometimes be highly unreliable. In particular, people
>giving free support can quite reasonably take the attitude that you should
>spend a lot of your time creating an absolutely clear example, which can be
>a very time consuing excercise.

Moreover, the emphasis in such support is to improve the product, not
solve the user's problem.  So if a report of a bug is made and the bug
has already been fixed in the development sources, the person is
either told that or does not receive a response since their bug report
is no longer "useful".  But if they were a customer of a commercial
maintainer of the software, they would have to receive a patch unless
it was possible to make that new version available to them.

Re: gnat-3.10

[Dale Pontius]

In article <m2yb88sqc2.fsf@devo.ridgenet.net>,
        Ronald Cole <ronald@ridgenet.net> writes:
>
> Please learn to follow a thread.  I posted early on that I reported
> this bug several months ago and got back a patch from Richard Kenner
> but that it didn't seem to do the trick.
>
Well sorry about that, but our newsserver has already purged the
beginning of this whole thing. But that doesn't change the facts
of the whole support issue. Is Richard Kenner from ACT? If so,
and you did get a patch from him, then you got a freebie.

As someone else said, some folks charge to even make a bug report.

Dale Pontius
(NOT speaking for IBM)

Re: gnat-3.10

[Arthur Schwarz]

I love this stuff. Makes me long for Kindergarten. 

I'm going to tell my wife that this is more interesting
than 'Soaps', evening or daytime. A serial in the making.
Real life in the raw. Wasted bandwidth.

Go get'em (generic) guys.

arthur schwarz
aschwarz@acm.org
schwarza@gdls.com

Re: gnat-3.10

[Ronald Cole]

dewar@merv.cs.nyu.edu (Robert Dewar) writes:
> We have a number of large customers on HPUX. They were using 3.09 when it
> was the current release, most of them have by now switched to 3.10 which is
> the current product release. As usual, this will be released publicly in
> the future. I must say that the logic that says "I, Ronald Cole, have failed
> in attempting to build from sources, therefore no other customers are using
> this successfully", entirely elludes me.

That's not what I said at all, Robert.

> As for ignorance. Yes, I am ignorant of your specific problem, because we
> have not looked at it. I simply deleted your long post. If indeed you think
> of it as a bug report, then bug reports get submitted to report@gnat.com.
> Posting them to CLA will not help get them fixed!

You can save yourself a lot of grief here, Robert, by reading what I
wrote and by not putting words into my mouth.

-- 
Forte International, P.O. Box 1412, Ridgecrest, CA  93556-1412
Ronald Cole <ronald@ridgenet.net>        Phone: (760) 499-9142
President, CEO                             Fax: (760) 499-9152
My PGP fingerprint: E9 A8 E3 68 61 88 EF 43  56 2B CE 3E E9 8F 3F 2B

Re: gnat-3.10

[Ronald Cole]

dewar@merv.cs.nyu.edu (Robert Dewar) writes:
> We have a number of large customers on HPUX. They were using 3.09 when it
> was the current release, most of them have by now switched to 3.10 which is
> the current product release. As usual, this will be released publicly in
> the future.

GNAT is based on GCC, right?  What section of the GPL are you relying
on to justify making binary releases of modified GPL software to your
paying clients and refusing to make the source code available to the
public at the same time?  Under Section 3 of the GPL, it seems clear
that once you distribute an object code work based on a GPL'd program,
you must release the source code concurrently.

And why hasn't the text of the GPL been included in your previous
source distributions?

-- 
Forte International, P.O. Box 1412, Ridgecrest, CA  93556-1412
Ronald Cole <ronald@ridgenet.net>        Phone: (760) 499-9142
President, CEO                             Fax: (760) 499-9152
My PGP fingerprint: E9 A8 E3 68 61 88 EF 43  56 2B CE 3E E9 8F 3F 2B

Re: gnat-3.10

[Larry Kilgallen]

In article <m2oh8ydzt8.fsf@devo.ridgenet.net>, Ronald Cole <ronald@ridgenet.net> writes:
> dewar@merv.cs.nyu.edu (Robert Dewar) writes:
>> We have a number of large customers on HPUX. They were using 3.09 when it
>> was the current release, most of them have by now switched to 3.10 which is
>> the current product release. As usual, this will be released publicly in
>> the future.
> 
> GNAT is based on GCC, right?  What section of the GPL are you relying
> on to justify making binary releases of modified GPL software to your
> paying clients and refusing to make the source code available to the
> public at the same time?  Under Section 3 of the GPL, it seems clear
> that once you distribute an object code work based on a GPL'd program,
> you must release the source code concurrently.

I thought the rule was merely that one had to distribute source to those
to whom one distributed the binary, without restriction on their right
to redistribute.

Certainly requiring an innovator to engage in distribution to other
individuals would be an onerous burden.  Since ACT chooses to use
the Internet for their distributions anyway they can make public
"distributions" cheaply, but the fact that they do so should not
burden them any more than any other innovator.

The general way of getting the sources would then be to find one of
those large customers who are paying for support.  I would advise a
gentler approach than has been used in this thread.

Larry Kilgallen

Re: gnat-3.10

[Robert Dewar]

Ronald Cole said

<<dewar@merv.cs.nyu.edu (Robert Dewar) writes:
> We have a number of large customers on HPUX. They were using 3.09 when it
> was the current release, most of them have by now switched to 3.10 which is
> the current product release. As usual, this will be released publicly in
> the future.

GNAT is based on GCC, right?  What section of the GPL are you relying
on to justify making binary releases of modified GPL software to your
paying clients and refusing to make the source code available to the
public at the same time?  Under Section 3 of the GPL, it seems clear
that once you distribute an object code work based on a GPL'd program,
you must release the source code concurrently.

And why hasn't the text of the GPL been included in your previous
source distributions?

--
Forte International, P.O. Box 1412, Ridgecrest, CA  93556-1412
Ronald Cole <ronald@ridgenet.net>        Phone: (760) 499-9142
President, CEO                             Fax: (760) 499-9152
My PGP fingerprint: E9 A8 E3 68 61 88 EF 43  56 2B CE 3E E9 8F 3F 2B
>>



Some confusions here. First there is no requirement that we release anything
at all to the public. Ronald you may want to look up the GPL yourself, it
has no such requirment at all. 

The requirement is that we make available sources on request to anyone
to whom we make objects available. Also there is no requirement to "release"
the source code, just to make it available at a reasonable copying charge.
(as I say, Ronald, you should carefully read the GPL, since you seem to have
some misconceptions about it).

However, we do in fact release binary versions, though we are not required to
do so, and in accordance with the GPL, we do indeed release sources for these
binary versions.

As to file COPYING being missing, I will investigate, for convenience it
should be in the release, although you can always get a copy as noted in
the headers.

We are strong supporters of the GPL, and carefully follow it, and insist that
all users of GNAT also follow it (this in practice only restricts people making
modificatoins to the compiler, since the runtime is released under a modified
GPL that places almost no restrictins in practice on its use).

We often find that people have strange misconceptions about the GPL, in
particular they think that you are required to distribute stuff, you are
not! and that you are required to distribute sources free, you are not.

The distinction is important, we like people to know that we take the extra
effort to make versions of GNAT public not because we have to, but because
we choose to, since we think it is valuable to the Ada community.

Robert Dewar
Ada Core Technologies

Re: gnat-3.10

[Corey Minyard]

Ronald Cole <ronald@ridgenet.net> writes:
> 
> dewar@merv.cs.nyu.edu (Robert Dewar) writes:
> > We have a number of large customers on HPUX. They were using 3.09 when it
> > was the current release, most of them have by now switched to 3.10 which is
> > the current product release. As usual, this will be released publicly in
> > the future.
> 
> GNAT is based on GCC, right?  What section of the GPL are you relying
> on to justify making binary releases of modified GPL software to your
> paying clients and refusing to make the source code available to the
> public at the same time?  Under Section 3 of the GPL, it seems clear
> that once you distribute an object code work based on a GPL'd program,
> you must release the source code concurrently.

The GPL never states that you have to make a public release or you
have to give anything to anybody.  It states that if you give someone
a binary release, you have to give them source (which ACT does, the
source is available to customers).  It also says that you cannot
restrict what someone does with what you give them (as long as they
comply with GPL), so ACT asks customers not to release the sources,
but it really cannot require them to.

The GPL is not an easy document to read and requires careful reading
to understand.  Don't make rash statements about GPL without carefully
reading it.  It states that:

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

    a) Accompany it with the complete corresponding machine-readable
    source code, which must be distributed under the terms of Sections
    1 and 2 above on a medium customarily used for software interchange; or,

    b) Accompany it with a written offer, valid for at least three
    years, to give any third party, for a charge no more than your
    cost of physically performing source distribution, a complete
    machine-readable copy of the corresponding source code, to be
    distributed under the terms of Sections 1 and 2 above on a medium
    customarily used for software interchange; or,

    c) Accompany it with the information you received as to the offer
    to distribute corresponding source code.  (This alternative is
    allowed only for noncommercial distribution and only if you
    received the program in object code or executable form with such
    an offer, in accord with Subsection b above.)

and ACT complies with a) since they supply code on the FTP site with
the binary executables.

What they do not seem to do, which disturbs me a little, is distribute
the GCC base they use to build their stuff on their FTP site.  Since
the patches they supply don't apply perfectly, they are obviously
using a patched GCC base and do not provide information about how to
get it.  But perhaps they are relying on b) above in the GPL for that
specific part, which would be perfectly fine legally.  I'd rather have
it on the FTP site, though.

> 
> And why hasn't the text of the GPL been included in your previous
> source distributions?
> 

Since it is a patch to GCC, it probably doesn't matter.  It might be
better to include it in the src/ada directory for the compiler sources
just in case, though.  The library sources are under a different
license, so they don't apply.

-- 
Corey Minyard               Internet:  minyard@acm.org
  Work: minyard@nortel.ca       UUCP:  minyard@wf-rch.cirr.com

Re: gnat-3.10

[Richard Kenner]

In article <m2oh8ydzt8.fsf@devo.ridgenet.net> Ronald Cole <ronald@ridgenet.net> writes:
>Under Section 3 of the GPL, it seems clear that once you distribute an
>object code work based on a GPL'd program, you must release the source
>code concurrently.

This is a common misconception.  See the FSF web pages for a more
detailed discussion of this issue.

In general, the GPL does not (and cannot) impose any requirement to
distribute anything at all, nor to make anything available in any way.

It imposes precisely two requirements: (1) to not put any restrictions
on redistribution and (2) to make sources available when something is
distributed in binary form.  The latter includes a requirement that
the fee for such distribution be limited to cost.  This is the only
requirement on limiting prices in the GPL and is because it would
otherwise be possible to effectively not provide the sources by
setting a very high price on them.

Here is Section 3 that you reference.  Note that there are options (a)
and (b) that would apply to your scenario.  You act as though (a) was
the only option.

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

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

    a) Accompany it with the complete corresponding machine-readable
    source code, which must be distributed under the terms of Sections
    1 and 2 above on a medium customarily used for software interchange; or,

    b) Accompany it with a written offer, valid for at least three
    years, to give any third party, for a charge no more than your
    cost of physically performing source distribution, a complete
    machine-readable copy of the corresponding source code, to be
    distributed under the terms of Sections 1 and 2 above on a medium
    customarily used for software interchange; or,

    c) Accompany it with the information you received as to the offer
    to distribute corresponding source code.  (This alternative is
    allowed only for noncommercial distribution and only if you
    received the program in object code or executable form with such
    an offer, in accord with Subsection b above.)

The source code for a work means the preferred form of the work for
making modifications to it.  For an executable work, complete source
code means all the source code for all modules it contains, plus any
associated interface definition files, plus the scripts used to
control compilation and installation of the executable.  However, as a
special exception, the source code distributed need not include
anything that is normally distributed (in either source or binary
form) with the major components (compiler, kernel, and so on) of the
operating system on which the executable runs, unless that component
itself accompanies the executable.

If distribution of executable or object code is made by offering
access to copy from a designated place, then offering equivalent
access to copy the source code from the same place counts as
distribution of the source code, even though third parties are not
compelled to copy the source along with the object code.

Re: gnat-3.10

[Dale Stanbrough]

Richard Kenner writes:

">So, if any part of my program contains any part (or derivation of)
 >their program, I have two choices:
 >1. Distribute my program "as a whole" under the terms of the GPL
 >2. Don't distribute my program.
 
 That's correct and exactly the status of the resulting work.  If you
 want to continue to view your code as proprietary, then you have two
 different copyright terms for pieces of the code and the only way to
 satisfy both is not to distribute the work at all."


...and there I think is a major issue that needs explaining (to me at
least!).
What does "contains" mean? If I write a GUI front end for Gnat, does that
constitute containment? I could distributed it separately, and have it just
make system calls to Gnat? What about a program that is linked to some of the
GNAT parser routines? What about a system that dynamically loads Gnat in as
a 
DLL? Is the calling convention what determines the "contains" relationship?


Dale ?:-/

Re: gnat-3.10

[Ronald Cole]

kilgallen@eisner.decus.org (Larry Kilgallen) writes:
> I thought the rule was merely that one had to distribute source to those
> to whom one distributed the binary, without restriction on their right
> to redistribute.

After re-reading the GNU Manifesto, it's clear that Stallman only
wants to make sure that one can get the source code to any GPL'd
binary one finds himself in possession of.  Even though Stallman pays
lip service to Kantian ethics ("Since I do not like the consequences
that result if everyone hoards information, I am required to consider
it wrong for one to do so.  Specifically, the desire to be rewarded
for one's creativity does not justify depriving the world in general
of all or part of that creativity." [GNU Manifesto]), it would appear
that the GPL is powerless to prevent the following scenario:

		  MAKE MONEY FAST HOARDING GPL'D SOFTWARE
			  Version 1, June 1997

1.  Make very useful enhancements to software that Stallman and others
    have worked hard on to make "free".
2.  GPL those useful enhancements.
3.  Find companies that are willing to pay big bucks for said
    enhancements.
4.  Said companies, after paying through the nose for GPL'd software,
    are unwilling to re-distribute because the GPL apparently only
    obligates one to distribute source if one distributes binaries.
    (Why would someone want to incur this obligation for something
    they had to pay a lot of money to acquire in the first place?)

Thus, as an implementation of the "Golden Rule", the GPL loses.

> The general way of getting the sources would then be to find one of
> those large customers who are paying for support.  I would advise a
> gentler approach than has been used in this thread.

Doubtful ACT will ever give out it's client list.  I know I wouldn't
give out mine.

-- 
Forte International, P.O. Box 1412, Ridgecrest, CA  93556-1412
Ronald Cole <ronald@ridgenet.net>        Phone: (760) 499-9142
President, CEO                             Fax: (760) 499-9152
My PGP fingerprint: E9 A8 E3 68 61 88 EF 43  56 2B CE 3E E9 8F 3F 2B

Re: gnat-3.10

[Larry Kilgallen]

In article <m2u3ijkf7d.fsf@devo.ridgenet.net>, Ronald Cole <ronald@ridgenet.net> writes:

> that the GPL is powerless to prevent the following scenario:

On further investigation, I find the GPL is also powerless
to prevent my car from running out of gasoline.

There have been 3 other software licensing schemes discussed,
and any of them could be adopted by someone writing software.

Feel free to adopt one and write an Ada compiler using those rules.
Reopen this topic when the compiler passes the validation tests.

Larry Kilgallen

Re: gnat-3.10

[Robert Dewar]

Ronald said

<<Doubtful ACT will ever give out it's client list.  I know I wouldn't
give out mine.
>>

Well certainly we won't give anyone a complete client list, since many of
our clients would not appreicate this kind of publicity. However, in cases
where we have permission from clients, we are generally happy to publicize
the fact that they are clients, and have often done so in the past.

We are otherwise highly protective of our clients interests, and respect
their wishes for not having their names revealed, as I am sure people
perfectly well understand.

Robert Dewar
Ada Core Technologies

Re: gnat-3.10

[Robert Dewar]

Ronald says

<<                  MAKE MONEY FAST HOARDING GPL'D SOFTWARE
                          Version 1, June 1997

1.  Make very useful enhancements to software that Stallman and others
    have worked hard on to make "free".
2.  GPL those useful enhancements.
3.  Find companies that are willing to pay big bucks for said
    enhancements.
4.  Said companies, after paying through the nose for GPL'd software,
    are unwilling to re-distribute because the GPL apparently only
    obligates one to distribute source if one distributes binaries.
    (Why would someone want to incur this obligation for something
    they had to pay a lot of money to acquire in the first place?)
>>


Well I guess Ronald finally understands the GPL :-)

In practice this scenario is of limited likelihood, and that is why it
has not happened in practice. If one charged "big bucks" for the enhancements
you had made, and if companies were willing to pay the big bucks, then an
obvious thing happens:

   1. If you charge say $1 million for your wonderfgul enhancements

   2. And lots of companies are willing to pay $1 million

   3. Then someone will buy it for $1 million

   4. And undercut your business by selling it for the bargain price of 
	say $200,000

But as I have said all along, there is nothing that requires you to 
distribute something you do under the GPL. One of the freedoms it confers
is the freedom to distribute your work to whomever you please. The
ultimate hoarding that can occur is if someone makes wonderful software
and keeps it to themselves completely, but no one in the GNU world thinks
for a monment that this should be prevented.

As I have frequently noted, there is nothing that requires ACT to continue
to make binary versions of GNAT available. It is something we choose to do as
a service to the Ada community.

Ron will just have to wait like everyone else for the public release of
3.10 which will happen in due time when we judge it to be appropriate. At
that time Ron can pick up a copy (whose availability is made possible by 
ACT customers who do pay for support :-) and start griping about 3.10 :-)

Robert Dewar
Ada Core Technologies

P.S. I quite realize that there are people with quite extreme views on
how software should be distributed and shared, but you should not assume
that everyone else shares these views. Now if Ron would put his energy
into providing free software, rather than trying to get hold of 3.10
before we consider it ready for public release, that would be more helpful.
But I guess the communistic view always appeals more to those who need what
they do not have, as opposed to those who have what they do not need :-)

Re: gnat-3.10

[Ronald Cole]

dewar@merv.cs.nyu.edu (Robert Dewar) writes:
> Well I guess Ronald finally understands the GPL :-)

Yes, Stallman left a hole in the GPL big enough for you to drive
your truck through and set up a toll gate.

> In practice this scenario is of limited likelihood, and that is why it
> has not happened in practice. If one charged "big bucks" for the enhancements
> you had made, and if companies were willing to pay the big bucks, then an
> obvious thing happens:
> 
>    1. If you charge say $1 million for your wonderfgul enhancements
> 
>    2. And lots of companies are willing to pay $1 million
> 
>    3. Then someone will buy it for $1 million
> 
>    4. And undercut your business by selling it for the bargain price of 
> 	say $200,000

This assumes that the company that paid big bugs for a solution is
willing to "burden" themselves with the GPL's redistribution
requirements in order to recoup their costs.  There are many other
less-painful ways to recoup costs that do not benefit free software,
and so I think it's likelihood isn't as limited as you appear to
think it is.

> But as I have said all along, there is nothing that requires you to 
> distribute something you do under the GPL. One of the freedoms it confers
> is the freedom to distribute your work to whomever you please. The
> ultimate hoarding that can occur is if someone makes wonderful software
> and keeps it to themselves completely, but no one in the GNU world thinks
> for a monment that this should be prevented.

It's your choice to distribute your work under the GPL.  Once one
chooses to distribute their work to person-A under the GPL, one should
not be allowed to say "no" if person-B asks for a copy.  That is why
Richard wrote the following: "Thus, there's no requirement to make a
public announcement of a source release to accompany the binary
release, but you may have to provide the sources to any number of
people other than those who got the binary directly from you, if they
request it."  This is just a reiteration of the Kantian philosophy
that is supposed to be embodied in the GPL.

All I would like is a provision in the GPL that prevents people from
doing what you did.  Perhaps I can convince Richard of the importance
of including his statement in the GPL before gcc-2.8 is released.

> Now if Ron would put his energy into providing free software, rather
> than trying to get hold of 3.10 before we consider it ready for
> public release, that would be more helpful.

You yourself said that 3.10 is your "current production release".  Do
your customers know that you consider that it's not even ready for
public release?  And why would I want to provide free software to the
world under a licensing agreement that makes it less-than-freely-
available to everyone?

> But I guess the communistic view always appeals more to those who
> need what they do not have, as opposed to those who have what they
> do not need :-)

It sure does.  That's why you are using the gcc backend instead of
writing one yourself.  The GPL loophole appeals to your sense of
greed.  If you really were to "walk the walk", you wouldn't say no to
my request for a copy because you would be making your money just
charging for support.  I guess you're afraid of free market
competition.

Besides, Communism forbids the ownership of property, so I think you
meant to say "socialistic view".  Regardless, use of the GPL on one's
work is voluntary and is thus libertarian.

-- 
Forte International, P.O. Box 1412, Ridgecrest, CA  93556-1412
Ronald Cole <ronald@ridgenet.net>        Phone: (760) 499-9142
President, CEO                             Fax: (760) 499-9152
My PGP fingerprint: E9 A8 E3 68 61 88 EF 43  56 2B CE 3E E9 8F 3F 2B

Re: gnat-3.10

[Robert Dewar]

Ronald says

<<You yourself said that 3.10 is your "current production release".  Do
your customers know that you consider that it's not even ready for
public release?  And why would I want to provide free software to the
world under a licensing agreement that makes it less-than-freely-
available to everyone?
>>

Right, when a production release is first made, it obviously has not been
as widely used as a production release that has been made some time ago
and widely used.

When we first make a prerelease, those customers that want to go to the
new version, to get the benefit of bug releases and new features can.
Some choose to, some choose not to. Those who choose to move to the new
release know that we will provide immediate help to them if there are 
any problems. This does not guarantee freedom from problems, but it
helps to reduce the risk.

For example, if there is a slight installation glitch that causes trouble
on some particular hardware configuration, we can easily help anyone
who runs into it, and let people know how to get around it. 

But a public release is a much different thing. Thousands will use it
who have no access to support from us or from any one else. It is therefore
much more important that public releases be glitch free, particularly
when it comes to installation problems.

That is why we delay making public releases until we think that all
the glitches have been ironed out. Some of our customers wait until
well after the public release to switch versions, which seems quite
reasonable, others want to move immediately to the new release.

Ronald is all upset over "what we have done to him", but all we have done
to him is to tell him to wait until we think that the version is ready for
public distribution. As we all know, Ronald had trouble with the 3.09
public release, perhaps that says we should have waited LONGER rathern
than SHORTER to distribute 3.09, though one cannot be sure that all
problems will be solved at any time.

Incidentally Richard Stallman is definitely supportive of the notion of
not releasing free software before it is in an appropriate state. The
Linux folks have often argued that the Linux distributes should contain
the latest GCC snapshots, since obviously they fix bugs, but such an
attitude is not a good idea in the long run.

Sure, everyone would like to be able to get the latest and greatest GNAT
as early as possible, but we do not think it will help the community to
have very frequent releases which have not been thoroughly user tested.

We are willing to let customers have versions earlier, precisely because
we know we are there to help if there are any problems. 

Actually Ronald's experience in failing to build from sources is a really
good illustration of the problem. If Ronald had been a customer, then of
course we would be spending time with him to get around the problem (whether
the problem is our problem or the customer's problem is not the issue in
this situation, getting around the problem *is* the only issue). On the
other hand, Ronald's frustration with 3.09 is a good example of what we
want to avoid happening by making sure that public releases are in good
shape before they occur.

This is simply a matter of quality control, which seems in our judgment
to be as important for public releases of GNAT as it is for any other
customer releases that we make.

Robert Dewar
Ada Core Technologies

Re: gnat-3.10

[Robert Dewar]

Ronald says

<<It sure does.  That's why you are using the gcc backend instead of
writing one yourself.  The GPL loophole appeals to your sense of
greed.  If you really were to "walk the walk", you wouldn't say no to
my request for a copy because you would be making your money just
charging for support.  I guess you're afraid of free market
competition.
>>


The reason we say no to your request for a copy of the latest version is
that we do not want you to waste time as you did on 3.09. Apparently, you
expected free support on those problems, and got a bit upset when we
declined. Since no free support will be available on 3.10 either, we
will not give you a copy until we feel it is in reasonable shape for
a public release.

Robert Dewar
Ada Core Technologies

Re: gnat-3.10

[David Kastrup]

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1: Type: text/plain, Size: 2014 bytes --]


Ronald Cole <ronald@ridgenet.net> writes:

> It's your choice to distribute your work under the GPL.  Once one
> chooses to distribute their work to person-A under the GPL, one
> should not be allowed to say "no" if person-B asks for a copy.


No, no, NOOO!  That way nobody will put anything under the GPL because
it would legally require him to set up a proper distribution service.
You can't demand that from freeware authors.

In fact, the GPL expressively frees the author of software from having
to hassle out the details of how to best distribute his software, and
leave that to others and in particular the market (it does not impose
any cost restrictions on this process).

This, of course, also means that if you're a lousy distributor (or
uploader), your software might not get anywhere worth noting.  But to
hold freeware authors responsible for their lack of resources or
management when they provide something for free is foolish.

Fortunately, the GPL does *not* demand that you're required to be a
proper distributor, *unless* you choose to make only binaries
available (in which case you are required to have the infrastructure
for providing the source).  If you don't distribute binaries without
source, you're complying to the GPL, regardless of how few people you
are distributing to.

The GPL serves just a few purposes: that no crippleware without source
gets effectively distributed (people should at least always get the
right to the source together with binaries), and that there are no
limits to redistribution other than that which would stop these
freedoms short.

I certainly would not want to have people go to the slammer because
they put something under the GPL and then their Internet connection
breaks down after they put out the announcement.


-- 
David Kastrup                                     Phone: +49-234-700-5570
Email: dak@neuroinformatik.ruhr-uni-bochum.de       Fax: +49-234-709-4209
Institut f�r Neuroinformatik, Universit�tsstr. 150, 44780 Bochum, Germany

Re: gnat-3.10

[Fergus Henderson]

Ronald Cole <ronald@ridgenet.net> writes:

 >dewar@merv.cs.nyu.edu (Robert Dewar) writes:
 >> If one charged "big bucks" for the enhancements
 >> you had made, and if companies were willing to pay the big bucks, then an
 >> obvious thing happens:
 >> 
 >>    1. If you charge say $1 million for your wonderfgul enhancements
 >> 
 >>    2. And lots of companies are willing to pay $1 million
 >> 
 >>    3. Then someone will buy it for $1 million
 >> 
 >>    4. And undercut your business by selling it for the bargain price of 
 >> 	say $200,000
 >
 >This assumes that the company that paid big bugs for a solution is
 >willing to "burden" themselves with the GPL's redistribution
 >requirements in order to recoup their costs.

No it doesn't.  They can still undercut your business by selling
your GPL'd software, with source, for $200,000.  They don't need to
distribute their own software at all, and they won't be "burdened".

--
Fergus Henderson <fjh@cs.mu.oz.au>   |  "I have always known that the pursuit
WWW: <http://www.cs.mu.oz.au/~fjh>   |  of excellence is a lethal habit"
PGP: finger fjh@128.250.37.3         |     -- the last words of T. S. Garp.

Re: gnat-3.10

[Ronald Cole]

dewar@merv.cs.nyu.edu (Robert Dewar) writes:
> The reason we say no to your request for a copy of the latest version is
> that we do not want you to waste time as you did on 3.09.

Well, this is certainly the first time you have said this!  Your public
release of 3.09 certainly was a waste of time on the HPUX platform...

> Apparently, you expected free support on those problems, and got a
> bit upset when we declined.

This is a lie.  All I did was report the optimizer bug to
report@gnat.com back in early February.  Richard Kenner kindly sent a
patch, but I didn't test it out because I thought (erroneously, with
the benefit of 20/20 hindsight) that you would probably make a "patch"
release (like are made for gcc or emacs when serious platform problems
are encountered) right away.

At no time did I demand that you fix it for me.

Last month, I had heard that some of your customers were enjoying 3.10
on HPUX and so I asked if 3.10 for HPUX could at least compile itself
according to your instructions.  When you rather-rudely indicated that
1) yes, they were, and 2) yes, it could, I asked you for the source.

> Since no free support will be available on 3.10 either, we will not
> give you a copy until we feel it is in reasonable shape for a public
> release.

Since you distributed 3.09 publicly in such bad shape for the HPUX
platform, I have reason to doubt that the public 3.10 will be any
different.  Additionally, I have reason to believe that your
definition of "reasonable" adheres to a standard that is quite a bit
lower than mine.

-- 
Forte International, P.O. Box 1412, Ridgecrest, CA  93556-1412
Ronald Cole <ronald@ridgenet.net>        Phone: (760) 499-9142
President, CEO                             Fax: (760) 499-9152
My PGP fingerprint: E9 A8 E3 68 61 88 EF 43  56 2B CE 3E E9 8F 3F 2B

Re: gnat-3.10

[James Rogers]

Just a slight detour from the current thread of this subject,
when will gnat-3.10 be generally available to the public on for
free?  How about gnat-4.0?

-- 
Jim Rogers
*************************************************************
Team Ada

Re: gnat-3.10

[Robert Dewar]

Cory says

<<The GPL never states that you have to make a public release or you
have to give anything to anybody.  It states that if you give someone
a binary release, you have to give them source (which ACT does, the
source is available to customers).  It also says that you cannot
restrict what someone does with what you give them (as long as they
comply with GPL), so ACT asks customers not to release the sources,
but it really cannot require them to.
>>

To clarify here, we tell customers that we think it is in the best
interests of GNAT if prereleases and wavefronts are not released
generally (for all the reasons I have previously stated), but that is
absolutely right, we cannot require it.

This is a similar situation to the GCC snapshots, which are available on
a limited basis, with a similar note that it is in the best interests
of the GNU project if these snapshots are not distributed. But there too,
it cannot be required, and in fact at least on one occasion, a public
release of Linux on CD/ROM has included the snapshot sources, much to
the consternation of FSF and Richard Stallman.

Richard Stallman is in fact quite insistent on the undesirability of
releasing snapshots generally. His concern is just like ours. It is
one thing to have a single user run into difficulties under conditions
where these difficulties can be immediately addressed. It is quite another

to have a public release where there are problems. In the latter case,
there will be general yelling and complaining, and this does not help

the GCC community at all.

I will reiterate that our release policies are geared to the needs of
the general GNAT consuming public, which is a very large set of people
at this stage, most of whom have very little expertise in getting arond
problems (you would be amazed at the number of reports of installation
problems we have that arise simply from not carefully following the
installation instructions).

Yes, there are a few enthusiasts who would like to fiddle with the
absolutely latest version, but we do not gear our public release
policies to this small group!

P.S. When I mention Richard Stallman's insistence on the undesirability
of releasing snapshots, I am reminded of an incident a while ago in which
we discussed with him the possibility of releasing snapshots with one
public version of GNAT. He got very agitated, and yelled at us (this
was in person, not by email), and we quite understood this reaction :-)

Note that the issue of figuring out appropriate release dates is one that
all vendors of software face. The fact that software is free software
really makes no difference at all. There are people who would love to get
their hands on Windows 97 right now, but Microsoft has decided it is not
ready yet.

Sure, vendors may make mistakes, but the temptation is in the direction
of releasing too early. Just the other day, I suggested an early public
release of 3.10 to the team here, but the discussion overruled the
suggestion, since people pointed out several points (including the issue
of dragging in the tasking stuff all the time) that really need solving
before the public release.

You should understand that I am eager to get 3.10 out as early as possible,
for one thing, I would like people to be able to use the super-neat SPITBOL
stuff that I have added. But it is a long-term mistake to let this
eagerness result in a premature release.

Robert Dewar
Ada Core Technologies

Re: gnat-3.10

[Robert Dewar]

Corey says

<<What they do not seem to do, which disturbs me a little, is distribute
the GCC base they use to build their stuff on their FTP site.  Since
the patches they supply don't apply perfectly, they are obviously
using a patched GCC base and do not provide information about how to
get it.  But perhaps they are relying on b) above in the GPL for that
specific part, which would be perfectly fine legally.  I'd rather have
it on the FTP site, though.
>>

Generally the patches we supply *should* apply cleanly to the base. That
is certainly the intention. When it is not the case, it is a glitch. Right
now, the patch situation is quite messy, because of the very large delta
between 2.7.1 and 2.8. Virtually all of our gcc patches are incorporated
into 2.8, but not all the 2.8 patches are incorporated into our patch
file, only the ones that Ada needs. Keeping this straight is not easy,
and that is especially true in the case of the many different targets
we support.

We are expecting gcc 2.8 to be released very soon, and this should
greatly simplify the issue of getting the right version of gcc sources
and patches (at least for a transitory moment, the GNAT patch file
will become empty or very close to it).

The current gcc-272.dif is nearly 6000 lines long (this file originall
started out as a very small file of a few critical patches). Furthermore,
things are more complicated now because of the appearence of the 2.7.2.1
release (this was a kind of semi-official release demanded by the Linux
community to fix one not very important bug that got blown out of
proportion), so now we need a gcc-2721.dif as well.

As I say, these problems should completely disappear when gcc 2.8 is
issued. One of the good things here is that all GNAT patches do get
reflected in the main GCC sources that FSF develops almost immediately
(that's because the developer of these patches, and the maintainer
of the FSF version of GCC are the same person :-)

Re: gnat-3.10

[Richard Kenner]

In article <5os9i9$o32$1@goanna.cs.rmit.edu.au> Dale Stanbrough <dale@goanna.cs.rmit.edu.au> writes:
>...and there I think is a major issue that needs explaining (to me at
>least!).
>What does "contains" mean? If I write a GUI front end for Gnat, does that
>constitute containment? I could distributed it separately, and have it just
>make system calls to Gnat? What about a program that is linked to some of the
>GNAT parser routines? What about a system that dynamically loads Gnat in as
>a DLL? Is the calling convention what determines the "contains" relationship?

This is indeed a somewhat vague area and, to some extent, purposely so.
The line is more likely to be drawn in terms of the extent to which the
parts work together and are both necessary to accomplish a goal than
in terms of the technical mechanisms used to make the calls between
the pieces.

For example, if somebody made a GUI for GNAT that relied only on its
external specifications and didn't take advantage of anything
particular about GNAT's internal structure, it probably would not be
considered a derived work unless it actually linked in GNAT code.

If you're thinking of doing something "on the fringe" here, you do indeed
have to ask the copyright holder about it.

Re: gnat-3.10

[Robert Dewar]

Dale asks

<<...and there I think is a major issue that needs explaining (to me at
least!).
What does "contains" mean? If I write a GUI front end for Gnat, does that
constitute containment? I could distributed it separately, and have it just
make system calls to Gnat? What about a program that is linked to some of the
GNAT parser routines? What about a system that dynamically loads Gnat in as
a
DLL? Is the calling convention what determines the "contains" relationship?
>>

First of all you need to look at the GPL itself, not some summary of it
on CLA, and interpet the language there. Generally, if two parts of the
same system are separate Unix processes, with relatively thin interfaces,
the requirements will be met. Linking to some parser routines is obviously
past the line and is not allowed, dynamically loading GNAT will generally
be fine, writing a GUI will certainly be fine.

As is the case with any legal document, you have to look at the contract
terms, and look at the specific case to make sure that you are not violating
the contract. If you are in doubt, you can consult your own lawyer, or you
can ask the copyright holder, who may or may not enlighten you (the holder
of the copyright is under no obligation to give you advance notice of what
might or might not violate the copyright, and may well not be willing to
give you such advance notice if it is a dubious case).

As an example of a dubious case, trying to make a compiler by having the
GNAT front end write a file containing the tree, and a separate proprietary
backend reading the tree is probably dubious, and Richard Stallman, as
representative of the copyright holder, has been unwilling to give an
opinion as to whether this does or does not violate the conditions of the
GPL (he would of course show more interest in the subject if anyone
actually did this!)

On the other hand, a GUI front end for GNAT, such as the many ones that
have already been created, seems absolutely fine for everyone, although
so far, most of these GUI's have themselves been free software, which is
certainly to be encouraged. 

How about a GUI for GDB? Well that's getting closer to the line, perhaps
it is past it, because then the GUI becomes a fundamental part of the
debugger, whereas a GUI for GNAT is not a fundamental part of the compiler
at least not usually.

If all this makes people a bit nervous about what is and what is not allowed
in attempts to use GPL'ed stuff as part of proprietary software, do not
expect us or Richard Stallman to get too upset! If it encourages the
production of more freely available software for the Ada community, that is a good thing. An example
of this happening is the vendor who will be creating a new Ada 95
cross compiler using the GNAT front end and their own previously
proprietary backend, which in its new incarnation will join the
ranks of free software products.

Two particular notes on all this with respect to GNAT.

First, the ASIS interface has been specially designed so that it is suitable
for generating third party products that need not be covered by the GPL. One
of the functions we see for ASIS is providing this kind of insulation.

Second, none of this discussion ever applies to programs *generated* by
GNAT, since all units in question there are covered not by the GPL, but
by the modified GPL that is specifically designed to allow such programs
to be used without imposing GPL distribution requirements.

It only applies to programs that actually want to use some of our
copyrighted code directly.

Re: gnat-3.10

[Richard Kenner]

In article <m2n2oiknxv.fsf@acm.org> Corey Minyard <minyard@acm.org> writes:
>What they do not seem to do, which disturbs me a little, is distribute
>the GCC base they use to build their stuff on their FTP site.  Since
>the patches they supply don't apply perfectly, they are obviously
>using a patched GCC base and do not provide information about how to
>get it.

No, the patch file applies to an unmodified GCC 2.7.2.  It is applied
to that base every night automatically.  Yes, some of the patches were
originally derived from different GCC version of GCC sources, so the
line numbers aren't exact, but those exact patches are applied to the
2.7.2 sources each night to do builds.

Re: gnat-3.10

[Robert Dewar]

Peter Herman said

<<it is quite frustrating to frequently read this subject line and at the
same time seeing ez2load freezed at 3.07 since January 1997    :-(
>>

There was no DOS 3.09, because we ran into a nasty signal problem that
we could not solve under DJGPP. We are hoping that there will be a 
DOS 3.10, and have some ideas about how to solve these problems,but
the DOS port at this stage has very low priority, since, not suprisingly
there is very little commercial interest in this port. Still, we will
do our best to make a 3.10 DOS version available.

Robert Dewar
Ada Core Technologies

Peter: machines that run Windows 95 can be bought for a song these 
days, how about upgrading that old DOS system :-)

Re: gnat-3.10

[Peter Hermann]

it is quite frustrating to frequently read this subject line and at the
same time seeing ez2load freezed at 3.07 since January 1997    :-(

Robert and the Bobs, please don't reply: we outdoor folks know the answer.

--
Peter Hermann  Tel:+49-711-685-3611 Fax:3758 ph@csv.ica.uni-stuttgart.de
Pfaffenwaldring 27, 70569 Stuttgart Uni Computeranwendungen
Team Ada: "C'mon people let the world begin" (Paul McCartney)

Re: gnat-3.10

[Robert Dewar]

James Rogers asks

<<Just a slight detour from the current thread of this subject,
when will gnat-3.10 be generally available to the public on for
free?  How about gnat-4.0?
>>

For GNAT 3.10, we never announce schedules for the public releases, and
we have no definite plans for the 3.10 release. Development is continuing
very rapidly, and basically we wait till there is a nice window of
opportunity in terms of this development cycle till we make the public
release. At this stahge, we will probably wait till the gdb support is
complete, since this seems to be going very well.

The designation 4.0x is reserved for the officially validated version.
TO be using an officially validated version, you must have a contractual
relationship with the holder of the validation certificate, or of courrse
you could do the validation yourself. Thus 4.0 will only be available
to supported customers. It is still free software, but transitive
distribution does not make much sense in this formal case, since the
recipient of a further distribution would NOT be using a formally
validated compiler, according to DoD rules.

If all you are interested in is getting a compiler that is technically
equivalent to the validated compiler, then you have nothing to worry
about. We will release public versions that are indeed equivalent to
the validated versions, and which, for the targets for which we are
validated (13 of them at this stage), will pass all the validation tests.

Indeed we hope that version 3.10 will meet this goal (we run the entire
ACVC suite every night on several different machines).

Robert B.K. Dewar
Ada Core Technologies

Re: gnat-3.10

[James Rogers]

Ronald Cole wrote:
> 
> dewar@merv.cs.nyu.edu (Robert Dewar) writes:
> > The reason we say no to your request for a copy of the latest version is
> > that we do not want you to waste time as you did on 3.09.
> 
> Well, this is certainly the first time you have said this!  Your public
> release of 3.09 certainly was a waste of time on the HPUX platform...
> 
> > Apparently, you expected free support on those problems, and got a
> > bit upset when we declined.
 
> > Since no free support will be available on 3.10 either, we will not
> > give you a copy until we feel it is in reasonable shape for a public
> > release.

The ACT WEB page certainly has some misleading information concerning
support.  On the URL www.gnat.com/ftp/html is the following:

Downloading GNAT

Current supported platforms and versions are as follows:

      GNAT 3.09 Alpha Digital Unix 4.0
      GNAT 3.09 HP HPUX 10.x
      GNAT 3.09 x86 Linux ELF
      GNAT 3.09 SGI IRIX 5.3/6.2
      GNAT 3.09 SNI SINIX 5.45
      GNAT 3.09 Sparc Solaris 2.3/2.4/2.5.1
      GNAT 3.09 Sparc SunOS 4.1.3
      GNAT 3.09 Windows NT/Windows 95
      GNAT 3.09 IBM OS/2 Warp
      GNAT 3.09 Powermac Machten
      GNAT 3.07 DOS 

Gnat is available from these FTP sites and mirrors. 

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

I understand that ACT means that support is for a fee.  Unfortunately,
this is not stated or implied on this WEB page.  Since the versions
listed are available for free download, it is easy and not 
unreasonable to assume that *some* level of support is available
to those who do not pay a fee.

Perhaps the WEB page should read:

Currently available versions are as follows:
(Support for these versions is available for a fee)


-- 
Jim Rogers
*************************************************************
Team Ada

Re: gnat-3.10

[Ronald Cole]

dewar@merv.cs.nyu.edu (Robert Dewar) writes:
> As we all know, Ronald had trouble with the 3.09
> public release, perhaps that says we should have waited LONGER rathern
> than SHORTER to distribute 3.09, though one cannot be sure that all
> problems will be solved at any time.

You put Richard Kenner's lastest patches in the publicly released
gcc-272.dif the day before you released it.  Clearly, they weren't
well tested.  So, if you aren't willing to build and test the public
binary releases before you release the source, I would prefer that you
didn't allow last-minute changes to your "well tested" prereleases as
a matter of policy.  Particularly, if you're not going to make
"platform" fix releases (like gcc-2.7.2.1.tar.gz and
emacs-19.34b.tar.gz were).  I'll even go so far as to say that you
could have just put a newer gcc-272.dif file on cs.nyu.edu with a note
explaining the problem!

Not quite a year ago, I was able to build a combined gcc-2.7.2,
gnat-3.07, and g77-0.5.18 that bootstrapped perfectly with -O on HPUX.
That is a quite a tribute to Richard Kenner and his work on gcc!  I
would hope to be able to do the same with gcc-2.7.2.2, gnat-3.10, and
g77-0.5.20.

I am willing to wait for a well-tested public 3.10, just like I'm
willing to wait for gcc-2.8.  I am very confident that when gcc-2.8.0
is finally released that it will bootstrap with -O2 on the HPUX
platform!

> Incidentally Richard Stallman is definitely supportive of the notion of
> not releasing free software before it is in an appropriate state. The
> Linux folks have often argued that the Linux distributes should contain
> the latest GCC snapshots, since obviously they fix bugs, but such an
> attitude is not a good idea in the long run.

Linux is very stable because public "snapshots" are made available
quite frequently for a lot of people to build and report problems they
encounter.  Linux development is truly a community project.  I
generally don't mess with the development releases until Linus declares
that he is freezing features and readying the next stable version, at
which point I lend a hand in making it very stable by building it on
as many different platforms/configurations as I am able to.

> other hand, Ronald's frustration with 3.09 is a good example of what we
> want to avoid happening by making sure that public releases are in good
> shape before they occur.

I'd have to agree that it was pretty foolish to make a public release
of 3.09 that depended on untested/unproven last-minute patches to gcc.
I'll reserve judgement on 3.10 until you release it publicly.

> This is simply a matter of quality control, which seems in our judgment
> to be as important for public releases of GNAT as it is for any other
> customer releases that we make.

Then you might want to seriously consider building and testing the
public binary releases before you release the source publicly.

-- 
Forte International, P.O. Box 1412, Ridgecrest, CA  93556-1412
Ronald Cole <ronald@ridgenet.net>        Phone: (760) 499-9142
President, CEO                             Fax: (760) 499-9152
My PGP fingerprint: E9 A8 E3 68 61 88 EF 43  56 2B CE 3E E9 8F 3F 2B

Re: gnat-3.10

[Philip Brashear]

Mr. Cole,

Please give it up.  This is getting VERY boring to the rest of the CLA
community.  Have you not noticed that you have little or no support in
this crusade?  There is a reason for this.

At the very least, please consider nettiquite, and turn down the volume.

Phil Brashear
An unsupported use of GNAT

Re: gnat-3.10

[Robert Dewar]

James Rogers said

<<The ACT WEB page certainly has some misleading information concerning
support.  On the URL www.gnat.com/ftp/html is the following:
>>


Right, I see how this could be misleading, we will eliminate this page
right away. Thanks.

Re: gnat-3.10

[Daniel R Risacher]

While I understand and sympathize completely with ACT's position, (as
stated by Robert) I suggest that the more commonly used model for GNU
software is to have separate "stable" and "bleeding edge" releases.

Examples of this are Linux and the GIMP.  You cannot set up Linux
2.1.43 without seeing many notices warning that it is a developmental,
kernel-hacker's version.  But you can download it just the same.  I
wouldn't expect anyone to provide support for 2.1.43 without being
paid for it.  

The GIMP is similar.  0.54 is billed as a "mostly-stable" version, but
developers (and patient users) have been playing with 0.99.x for a
while.  Support for 0.99 consists only of a mailing list of other
users/developers.

Guile has automatic, daily snapshots of the latest code available.

I suggest this sort of policy might be a reasonable model for gnat
distribution.  I don't know; I don't use gnat, or Ada, for that
matter.

$2E-2, 
 Dan Risacher


>>>>> "Robert" == Robert Dewar <dewar@merv.cs.nyu.edu> writes:

    Robert> That is why we delay making public releases until we think
    Robert> that all the glitches have been ironed out. Some of our
    Robert> customers wait until well after the public release to
    Robert> switch versions, which seems quite reasonable, others want
    Robert> to move immediately to the new release.

    Robert> Robert Dewar Ada Core Technologies

Re: gnat-3.10

[Joerg Rodemann]

Robert Dewar (dewar@merv.cs.nyu.edu) wrote:
> Ronald says

> Incidentally Richard Stallman is definitely supportive of the notion of
> not releasing free software before it is in an appropriate state. The
> Linux folks have often argued that the Linux distributes should contain
> the latest GCC snapshots, since obviously they fix bugs, but such an
> attitude is not a good idea in the long run.

> This is simply a matter of quality control, which seems in our judgment
> to be as important for public releases of GNAT as it is for any other
> customer releases that we make.

In my opinion this is a very good point you made. Especially as Ada is ---
at least in our mindset --- strongly connected with Software Quality or
Quality Software. Certainly with Ada you could do the same crippled type
of software one often finds in projects using C, C++ or else. At least the
end users often seem not to care about anything like quality. Although they
mutter if somethings goes wrong or worse (system crash, harddisk overwrite).

So I believe it serves the Ada community very well if you provide us (and I'd
like to thank you for making it publicly available to us) with a solid version
of GNAT. this most probably will prohibit opinions and sensations that "Ada
is not useful" just because the compiler is not good work. Surely we often
hear these beliefs about Ada-83 in this group. As to the latest releases of
any software my favourite route is: if you really need a new feature or you
desire it like hell, try out the new ones. Otherwise stick to the running
version until another one seems to be of the same quality. A collegue of mine
is just fiddling around with his systems just because software X requires 
the latest release from Y which only runs in combination with the newest
--- and totally incompatible --- release of Z. It gets really annoying if
Y or Z are essential parts of a system like the OS itself or the development
system.

Greetings

Joerg


--
rodemann@mathematik.uni-ulm.de | Dipl.-Phys. Joerg S. Rodemann
Phone: ++49-(0)711-5090670     | Flurstrasse 21, D-70372 Stuttgart, Germany
-----------------------------+-----------------------------------------------
rodemann@rus.uni-stuttgart.de  | University of Stuttgart, Computing Center
Phone: ++49-(0)711-685-5815    | Visualization Department, Office: 0.304
Fax:   ++49-(0)711-678-7626    | Allmandring 30a, D-70550 Stuttgart, Germany

Re: gnat-3.10

[David Kastrup]

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1: Type: text/plain, Size: 2687 bytes --]


Daniel R Risacher <risacher@worldnet.att.net> writes:

> While I understand and sympathize completely with ACT's position,
> (as stated by Robert) I suggest that the more commonly used model
> for GNU software is to have separate "stable" and "bleeding edge"
> releases.

No, this is rather uncommon, except for few examples.  In fact, *all*
of the GNU software distributed by the FSF tends to be distributed
mostly in what is considered a stable version (although typically some
earlier versions are kept for safety).  There is no explicit
distinction into stable and development versions for public releases.

>  Examples of this are Linux and the GIMP.  You cannot set up Linux
> 2.1.43 without seeing many notices warning that it is a
> developmental, kernel-hacker's version.  But you can download it
> just the same.  I wouldn't expect anyone to provide support for
> 2.1.43 without being paid for it.

Depends on what you call "support".  Actually, most of the support and
development in the various Linux usenet groups focuses around the
development versions, as there the most remains to be done.  If you
mean installation support by major Linux distributors, they will
obviously support what they provide, and that will indeed typically be
a "stable" version.

>  The GIMP is similar.  0.54 is billed as a "mostly-stable" version,
> but developers (and patient users) have been playing with 0.99.x for
> a while.  Support for 0.99 consists only of a mailing list of other
> users/developers.

And what superior support is offered for 0.54?

>  Guile has automatic, daily snapshots of the latest code available.
>  I suggest this sort of policy might be a reasonable model for gnat
> distribution.  I don't know; I don't use gnat, or Ada, for that
> matter.

It is usually quite a hassle to provide compilable, useful source
every day.  At the very least, you have to use something like CVS for
version control.  For some smaller projects this trouble might not at
all be worth the cost.

There are added disadvantages if you give development versions to the
public:  you have to sort out bug information and user foolishnesses
from sometimes inscrutible bug reports, and this can take more time
than you want to invest in it.  If you do alpha testing (giving out
preliminary versions to selected testers only), you can sometimes be
more productive.

So, in short, I think that there are different models that can be used
with justification.


-- 
David Kastrup                                     Phone: +49-234-700-5570
Email: dak@neuroinformatik.ruhr-uni-bochum.de       Fax: +49-234-709-4209
Institut f�r Neuroinformatik, Universit�tsstr. 150, 44780 Bochum, Germany

Re: gnat-3.10

[Robert Dewar]

Just in case Ronald's latest post causes any further confusion, just let
me repeat the point that anytime you see a binary GNAT release from ACT,
it has *always* bootstrapped itself, since that is the way we build releases.
RC may or may not succeed in repeating this process for HPUX 3.10, but certainly we have 
succeeded in doing the bootstrap at ACT.

Re: gnat-3.10

[Ronald Cole]

dewar@merv.cs.nyu.edu (Robert Dewar) writes:
> Just in case Ronald's latest post causes any further confusion, just let
> me repeat the point that anytime you see a binary GNAT release from ACT,
> it has *always* bootstrapped itself, since that is the way we build releases.
> RC may or may not succeed in repeating this process for HPUX 3.10, but certainly we have 
> succeeded in doing the bootstrap at ACT.

Yes, but should this latest rehash cause any *further* confusion, it
has been demonstrated that the binary GNAT release for the HP PA-RISC
is unable to compile two files in the GNAT source release with *any*
level of optimization.

This, Robert claims, was a release that, as a matter of policy, ACT
held up until it was well-tested by customers "to ensure a quality
product".

All this demonstrates a couple of things:

1) customer driven testing is no substitute for quality/robustness
   testing.
2) the person who bootstrapped the port didn't deem the compiler
   crashes during the bootstrap serious enough to hold up releasing
   GNAT-3.09 publicly.
3) Robert is unafraid to bundle last minute, unverified, gcc patches into
   his public releases and undo any stated benefits for unsupported
   gnat users.

-- 
Forte International, P.O. Box 1412, Ridgecrest, CA  93556-1412
Ronald Cole <ronald@ridgenet.net>        Phone: (760) 499-9142
President, CEO                             Fax: (760) 499-9152
My PGP fingerprint: E9 A8 E3 68 61 88 EF 43  56 2B CE 3E E9 8F 3F 2B

Re: gnat-3.10

[Richard Kenner]

In article <m22056txii.fsf@devo.ridgenet.net> Ronald Cole <ronald@ridgenet.net> writes:
>Yes, but should this latest rehash cause any *further* confusion, it
>has been demonstrated that the binary GNAT release for the HP PA-RISC
>is unable to compile two files in the GNAT source release with *any*
>level of optimization.

That's simply not possible since the binary release was constructed by
compiling all the sources files.  There therefore must have been *some*
set of options that compiled them.