Re: GNAT extensions, recompilation (was Re: what DOES the GPL...)

Re: GNAT extensions, recompilation (was Re: what DOES the GPL...)

[Richard Kenner]

In article <5tp162$rg9$1@trumpet.uni-mannheim.de> mw@ipx2.rz.uni-mannheim.de (Marc Wachowitz) writes:
>Robert Dewar wrote:
>> Actually it would be very nice if there was more experimental activity
>> with GNAT in terms of interesting new experimental ideas, but so far
>> we haven't seen much of this -- hopefully we will see more in the future,
>> this is after all why the sources are available
>
>If my memory doesn't fool me, and I didn't misunderstand what you meant,
>you've recently stated (here or on chat) that you wouldn't consider the
>ability of non-customers to recompile the public releases very important,
>since you'd expect people to use the binary releases.
>
>Above you've already mentioned one yourself: Unless everyone trying to extend
>(or in some other way experiment with the internals of) GNAT is a customer,
>there should be a reasonable confidence that one will be able to re-apply
>one's modifications to further releases (obviously, with changes necessary
>for the new release). 

Right, but the point is that anybody with enough knowlege to usefully
do work in experimenting with GNAT extensions will have had to have
formed enough knowlege to know how to properly compile it, which, as
you know, is only a straighforward process if you understand what's
going on.

>Another reason is that there may be patches to GCC which one considers very
>important for one's platform, and which one therefore wants to be part of the
>GNAT version one is using, without depending on updates from whoever provides
>the binary releases.

This is actually a good reason *not* to support the building of GNAT
from sources since there is no assurance that those "important patches
to GCC" will work properly with GNAT.  There are quite a number of
patch sets to GCC out there.  Some will work with GNAT and some will
likely not, but there is nobody who reliably knows which are in which
category.  Somebody who tries something like this is most assuredly on
their own!

Re: GNAT extensions, recompilation (was Re: what DOES the GPL...)

[Marc Wachowitz]

Richard Kenner wrote:
> This is actually a good reason *not* to support the building of GNAT
> from sources since there is no assurance that those "important patches
> to GCC" will work properly with GNAT.

Obviously there isn't any assurance, but the decision whether such changes
are worth the risk is clearly to be made by whoever wants to apply patches,
and is reasonably confident to make at least an educated guess about this.

> Somebody who tries something like this is most assuredly on their own!

Of course; if I'd ever find what looks like a GNAT bug using such a patched
system, I'd certainly check whether it does also occur with the original
distribution, and not blindly report it to innocent developers. This should
go without saying, but I guess with users who don't even glimpse at system
documentation, you've probably minimized what you're willing to assume.

-- Marc Wachowitz <mw@ipx2.rz.uni-mannheim.de>

Re: GNAT extensions, recompilation (was Re: what DOES the GPL...)

[Richard Kenner]

In article <5tquu7$djj$1@trumpet.uni-mannheim.de> mw@ipx2.rz.uni-mannheim.de (Marc Wachowitz) writes:
>Of course; if I'd ever find what looks like a GNAT bug using such a patched
>system, I'd certainly check whether it does also occur with the original
>distribution, and not blindly report it to innocent developers. This should
>go without saying, but I guess with users who don't even glimpse at system
>documentation, you've probably minimized what you're willing to assume.

One might indeed hope it would go without saying, but the number of
bug reports sent to the GCC bug list that were based on people using
highly experimental patches from "random" sources strongly suggests
that this is indeed something that it should not be assumed people know.

Re: GNAT extensions, recompilation (was Re: what DOES the GPL...)

[Ronald Cole]

dewar@merv.cs.nyu.edu (Robert Dewar) writes:
> is about how to compile from sources -- indeed during the period when Ronald
> Cole was furiously complaining that he could not manage to compile the
> HP version from sources, there was an exchange on chat@gnat.com where
> one person who had successfully done this helped someone else to
> sucessfully rebuild the system.

Robert persists in implying that I was unable to build due to some
defect in my understanding of how to actually compile GNAT.  I have
been successfully building GNAT since version 2.04.  My "furious"
complaint was actually that GNAT (3.09) couldn't compile GNAT: due to
the introduction of a buggy fix to the patch file one is required to
apply to gcc before attempting compilation.

-- 
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 extensions, recompilation (was Re: what DOES the GPL...)

[Joerg Rodemann]

Ronald Cole (ronald@devo.ridgecrest.ca.us) wrote:
> dewar@merv.cs.nyu.edu (Robert Dewar) writes:
> > is about how to compile from sources -- indeed during the period when
> > Ronald
> > Cole was furiously complaining that he could not manage to compile the
> > HP version from sources, there was an exchange on chat@gnat.com where
> > one person who had successfully done this helped someone else to
> > sucessfully rebuild the system.

> Robert persists in implying that I was unable to build due to some
> defect in my understanding of how to actually compile GNAT.  I have
> been successfully building GNAT since version 2.04.  My "furious"
> complaint was actually that GNAT (3.09) couldn't compile GNAT: due to
> the introduction of a buggy fix to the patch file one is required to
> apply to gcc before attempting compilation.

NOOOOOO! Not again...please. I think we have talked this over already far too
much.

Yours

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