🚀 go-pugleaf

Sorry to everyone -- I promised to get a status report out "a few days from
now" a couple of weeks ago.  Some interesting things came up which have not
yet been resolved.

Here is a quickie status report -- I'm flying out of town in a few hours and
won't be back until 4/13 and want to get something out.

Thanks to Cygnus Support for letting me use a workstation and free parking
pass for almost 3 weeks in March, I finished pretty much all the "heavy
lifting" for GNU Fortran -- at least as far as technical issues.  There are
a number of little things to do, and one major thing: get COMPLEX support
in the GNU back end working.  I call this latter thing "major" because it
is uncertain how hard it will be.

Aside from COMPLEX and a few minor things that probably will take no more than
a day or a few hours each, GNU Fortran is ready for me to start testing on
my own before I inflict it on alpha testers.  However, I'm back at home with
only 2400-baud access (inadequate for compiler development IMHO) so I haven't
gotten any work done since losing access to the Sun-4 at Cygnus.  (I sort
of still have it, but not on a sufficiently full-time basis, and parking in
Cambridge is expensive; it's more of a last resort when needed.)

After I return from my short trip, I should be able to get a high-speed modem
and access to a high-speed modem at Cygnus during much of the day to continue
my work.  I'm hoping the phone-line quality in my local town is up to the
job.  Meanwhile, I'm planning on buying a major 486 system and getting Linux
(the new free GPL-protected UNIX-like OS for 386/486 systems) up and running.
Either way, I expect to be able to get back to GNU Fortran development at
least on a half-time basis pretty soon.

As my last status report said, I had hoped to alpha-test GNU Fortran starting
January 1, 1992.  That was because I was starting a full-time contract for
six weeks as of that date.  I narrowly missed finishing the technical stuff,
but probably (hard to tell at this point) widely missed the administrivia
needed to do the release.  The time at Cygnus was all I needed to put in the
final major features (integrating COMMON, EQUIVALENCE, DATA, plus other
small stuff I forget at the moment).  At the moment, I have no plans to do
contract programming, so my calendar is open.

There is, however, something important that came up to which I alluded at
the top of this post.  I don't want to go into details yet, because I am
still thinking about how to address the issue (probably by asking you all
for help in some way), but basically the question is, how to properly
release GNU Fortran so it is adequately supported.  (Basically, I want to
avoid using a syringe to release something that has a fire-hydrant level
of pressure waiting to be applied; Fortran's audience is vastly larger than
for C, C++, and pretty much every other language except Cobol.)  My plans
have been to do the necessary integration work on my own code (the front end
and integration code to the back end) but not necessarily the back end, and I
need to make sure that there is adequate centralized support for this, among
other things.  It might take me another few weeks to nail this down, and I
want to do so before starting alpha test.  I'm doing this because I think
the Fortran community deserves the best support possible for a free-software
Fortran compiler, and I know I cannot provide that support singlehandedly.

In the meantime, please:

DO NOT send me email telling me you want to be an alpha tester -- it might
turn out that I'm not even involved in picking the alpha testers (as I am
not well equipped to support the effort from an administrative point of view).

DO NOT ask me when alpha or beta test will happen -- I wildly missed my
guess at 1/1/92 with little excuse, so I've joined the rest of the FSF
staff and volunteers in avoiding saying anything about planned release
dates.

DO NOT ask me what features will be supported at first: beyond saying "similar
to AT&T's public-domain f2c program using gcc to compile the generated code,
but as one compiler and hopefully faster", I cannot say much useful on this
topic until alpha happens (along with whatever documentation goes with it).

DO offer to help me with funding (donations welcome) or equipment (loans
of high-end UNIX workstations preferred; outright donations okay, but after
initial porting I might not make adequate use of them), as it could make
a huge difference (especially if funding means I don't have to take a
full-time contract within a few months).

DO offer to hire me to continue GNU Fortran development, either here at home,
or (if the pay is good enough) at your site, for some stretch of time.  This
would be a good way to make sure _your_ needs were addressed sooner rather
than, possibly, later, in addition to helping everyone get a free compiler
that much earlier.

Well, I hope this satisfies y'all for a while!  I know it isn't much, and
perhaps not what you want to hear, but I have to pack for my trip.  Oh, the
good news: while adding and integrating various features, and while finding
and fixing bugs, I have been pleased that my code has remained quite easy
for me to work with.  As it weighs in at over 130,000 lines of code, I can't
believe this is due to the "usual reasons" one's own code is easy for that
one to deal with.  I just hope this turns out to be the case for other people
experienced in compiler development once it's released.
---

James Craig Burley, Software Craftsperson    burley@gnu.ai.mit.edu
Member of the League for Programming Freedom (LPF)