Re: Where is the Ada for LINUX Team site?

[Robert Dewar <robert_dewar@my-deja.com>]

In article <8v9vgk$v8j$06$1@news.t-online.com>,
  "Juergen Pfeifer" <juergen.pfeifer@gmx.net> wrote:

> But it has been brought to my attention that ACT and the FSF
> see major deficiencies in the ALT packages.

A little background here. Richard Stallman questioned why ACT
was not providing RPM's directly. He worried that our failure to
provide RPM's meant we were neglecting the needs of GNU/Linux.
I explained that there were two problems.

  1) There were some technical problems with shared libraries.
  The ALT folks felt that they were not severe enough, and
  decided that shared libraries were valuable enough not to
  worry too much about the problems.

  2) Precise Ada compliance requires the provision of the
  alternative FSU threads package, since Linux threads are
  not Annex D compliant. The ALT folks decided that for
  simple use, that did not matter.

THe current status is that problem 1) has now been solved in
the latest GNAT technology, but problem 2) remains.

I think here at ACT we quite understand decision 2) above, and
it seems just fine to provide these RPM's with limited
capability for the purposes for which they were being provided,
but I explained to Richard that we need at ACT to be
distributing a fully compliant version.

We will study the issues of providing RPM's with both
threads packages, and we think it can probably be done for
the next release of GNAT now that problem 1) is solved (problem
1 was a show stopper at previous points for ACT provision of
RPM's).

But to be absolutely clear, we always felt that the RPM's
provided at the ALT site have been very useful to a lot of
people, and have always pointed users of the public version
of GNAT in that direction. I know of no post from the FSF
that could even possibly be construed as saying that there
were major deficiencies in the ALT packaging. I have to wonder
if Juergen actually read the thread in question. If anyone was
being criticized in this thread it was ACT for *not* providing
these RPM's :-)

> All these discussions have been done behind my back
> without involving me. There seems to be an ongoing discussion
> of GNAT packaging issues where ALT is not involved

These discussions were on the main gnu mailing list, open to
participation by any interested parties (this is a very large
list, with lots of people being involved), "behind my back"
is a bit of an odd description of discussions happening on
the GNU list, given you decided not to participate in this
list, which is the main place that issues of this kind are
discussed.

> any my mail offering to transfer our packaging
> knowledge remained unanswered.

Well I must say I am a bit non-plussed by this! I received
a friendly message from Juergen on last Monday morning. I
have not answered yet, because last week was SigAda, and
a lot of email is backed up from being away. In that message
he offers to help us with RPM technology, and indeed that
will be very helpful.

> As I'm doing this as hobby the least thing I'm interested in
> is politics, flaming and FUD.

I really did not see anything vaguely like politics, flaming
or FUD on the list, so I am a bit of a loss here.

> Withdrawing my contribution is the easiest way for me to step
> out of the fireline.

Well of course anyone can withdraw their contribution, but I
definitely fail to see any fireline here (anyone who likes is
welcome to checkout the threads in question -- actually I think
it is quite healthy that active discussions of GNAT and how it
should be handled in the framework of the standard gnu/gcc
distributions is very healthy!) Juergen, I think you should
read the thread in question, and I think you will find that
it has been mischaracterized to you!

> It has also the advantage that people can now start from
> scratch without a need to take care of an existing and
> established packaging scheme.

I am not sure of what "take care of" here means. Most certainly
we won't start from scratch in the sense of ignoring what has
already been done. At the same time, we do need to figure out
how to deal with the issue of the multiple thread libraries.
Although most people couldn't care less, we care that the Ada
compiler that is part of the GNU/Linux system should indeed be
fully conforming to the standard.

Hopefully once the transition to the new public tree (which will
of course be 2.9x/3.0 based) is complete, many of these problems
will disappear. We are also hoping that Linux Threads can
be fixed to solve the scheduling problems

Robert Dewar
Ada Core Technologies


Sent via Deja.com http://www.deja.com/
Before you buy.