Re: gnat-3.10

[Ronald Cole <ronald@ridgenet.net>]

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