Re: Why not gnat Ada in gcc?

[Jeffrey A Law <law at cygnus dot com>]

  In message <20001102035442.3BA9F34D82@nile.gnat.com>you write:
  > No one said anything of the kind. What we said is that we cannot possibly
  > deal with the kind of instability that we see daily in the open gcc tree
  > as the base for the version we provide to customers.
OK.  Then do you see how to separate your needs as a company from the needs
of the GNU development community and how to resolve those differences in such
a way as to put everyone on equal footing in regards to GNU Ada?  I don't
think so, partly due to this lengthy discussion and partly because you've
never really tried to play this game in the past.

  > And I must say I
  > am equally dumbfounded by the claim that Cygnus has always provided to
  > its customers the current open tree, that's certainly not my understanding!
We've never done that.  Sorry if you got that impression.

We always have a certain amount of ongoing development that we provide to
our customers, then make available to the net.

  > Noone is saying that the internal ACT build is "the official FSF GNU Ada",
  > it is simply what we provide for our customers. Indeed one important
  > service we provide our customers is precisely a guarantee that they
  > get a very carefully quality assured version of the compiler, even if
  > they are getting a daily wavefront build.
Then can you see that if the ACT tree isn't the official GNU Ada tree
that the official FSF GNU Ada has to be somewhere else?  And that somehow
you have to manage the process of dealing with changes from multiple sources?
And that the burden for managing those changes has to fall on ACT, not the
other developers?

ie, let's assume that for some reason I need to make a series of changes to
the GNU Ada sources due to changes elsewhere in the compiler.  It will be
ACT's responsibility to make sure those changes get into ACT's repository
and that ACT does not clobber those changes when ACT wants to install some
of their changes into the GNU repository?

  > Patches and major modifications to the open tree will come from whoever
  > develops such changes. For the immediate future, most such changes will
  > clearly come from ACT, if and when others contribute, changes will
  > come from them.
Right.  Good.  I think we're agreed on this, but then again, I don't think
we ever disagreed about this.

Where I think we disagree is on the topic of how changes get into the
official GNU tree and how we're going to ensure that all developers are
on equal footing.

  > Yes, certainly there may be cases where significant new developments
  > are done in increments that make sense (the sudden appearence of the
  > ia32 port was after all a spectacular case of that happening for gcc,
  > I would certainly hope that we would have *nothing* like that happeniung
  > in the GNAT case, it seems VERY unfortunate to me to get major out-of-the-
  > blue changes of this kind, and I think that should be avoided in future.
What makes you think the ia32 port came out of the blue?  Oh yea, you haven't
been active in gcc discussions until the last month or so.  You have to rely
on what others told you.  Why don't you go back and see the archives of the
development list where you'll find significant discussions about the new
backend and collaborative work from various sources before it was installed
to the head of the GCC tree.


jeff