Re: Endian[n]ess, language-specific problems and help with egcs.

[Craig Burley <burley at gnu dot org>]

>>>>>> "Toon" == Toon Moene <toon@moene.indiv.nluug.nl> writes:
>
> Toon> 1. As soon as the term "fortran" or "g77" occurs, PLEASE REFRAIN
> Toon>    FROM REPLYING.  Craig, Dave, Kate, Mumit or myself will guide
> Toon>    the question(er) in providing more information to fortran@gnu.org
> Toon>    the mailing list meant to deal with GNU Fortran questions.
>
>I'd agree, but that's what Craig told me _not_ to do recently and
>g77.info now says to send questions to this list, not to fortran@gnu.
>I shall probably refrain too and try to write some more doc for people
>to ignore.

I should explain this a bit more here, though I'm still trying to
figure out just how all this "should" work, so I'm definitely open
to differing opinions.

While g77 expertise currently seems to "reside" mostly at the
<fortran@gnu.org> and <g77-alpha@gnu.org> addresses, there are
a number of deficiencies we haven't yet addressed (mostly to do
with stuff on my end) but that the egcs project has.  E.g. better
mailing-list management, WWW presence, mail-list archives, etc.

And we've always had the problem that we couldn't really help
with bugs that really seemed to be non-g77-probably-gcc bugs,
sometimes forwarding them to bug-gcc, sometimes asking the submitter
to try there, etc.  That is, there's no "seamless" transition of
a classic g77 bug report to "probably more pertinent to gcc" status.

Also, egcs has already brought g77 "into the fold", and this has
benefitted g77 (primarily, but not exclusively, the egcs version)
substantially, because it has allowed non-"g77 people" to join
the fun of making g77 better, at least indirectly (by improving the
back end).  E.g. HJ Lu, Richard Henderson, and others have
contributed worthwhile patches, but aren't on the g77-specific
mailing addresses maintained by me at the FSF.  (I don't maintain
info-gnu-fortran, FWIW; it's not relevant to this.)

Earlier, I used to feel g77 issues were too "specialized" to bring
up on the egcs lists, although I did try to stick with the
straightforward uses of the egcs mailing lists in the g77 docs
for reporting bugs and such.

E.g. I'd kind of "cringe" when seeing really narrow g77/Fortran
questions, and subsequent discussion, posted to the entire egcs
(or egcs-bugs) list.  The endianness thread would have been a
classic example.

But lately I've thought it might be better to let the egcs community
go ahead and see what kinds of issues come up regarding g77.  That
might mean lots of irrelevant stuff ends up in lots of uninterested
peoples' mailboxes, but we can't predict what's irrelevant and/or
who is uninterested ahead of time.

Most telling is that the amount of C++-specific material on these
lists *vastly* exceeds the Fortran-specific material.  Yet, while
I delete C++ items quite quickly, I *do* at least skim them a
bit to get a feel for the kinds of questions people ask "over
there".

So, unless it becomes clear there's a real need to separate out
g77 issues from the pack, I'd like to stick with the way we're
telling people to handle egcs-g77 bug reports -- by posting them
to egcs-bugs.  Same for general issues, patches, and so on.

That's basically what I'd asked Dave Love to do awhile back
(though apparently I didn't cc the fortran list itself, which
was probably a mistake).

And, <fortran@gnu.org> can remain, for the time being, the mailing
address for FSF-gcc-g77 questions.  (There's only a handful of
people on that list.)

Let me know if I'm missing something here.  I'm not too certain
(or opinionated) about any of the above.  It just seems more
of a "natural" way to do things within the egcs project.

        tq vm, (burley)