Subject: Re: the next release
To: Danny Thomas <D.Thomas@vthrc.uq.edu.au>
From: Chris G. Demetriou <email@example.com>
Date: 07/26/1994 22:23:17
> I believe that was the "deadline" agreed to with USL but is the core team
> happy that it really will be genuinely production quality release by then?
I think the answer is 'yes'. We've been finding 'loose ends' for
a while now, and are pretty much done getting them taped down.
(duct tape is a hacker's friend. 8-)
My guess on when 1.0 will ship is "just before 8/3, or just after 8/9."
> For all architectures? (I don't really expect them to be equally good,
> but "just the facts, maam")
Not all. "lots." A notable couple probably won't be "production
quality." Anyway, 1.0 is expected to include "production quality"
binaries for (in alphabetical order):
> who's glad he didn't have volunteer for the hassles in keeping up with
> current on his production 0.9 systems, particularly since no information
> seems to have been made public about goals for 1.0, approx release
> schedules (until cgd's announcment a few months ago), etc.
We were actually gearing up to do another release back around
February. However, while that was in progess, i received a Very Nice
Letter from one Burt Levine, whose job title happens to be "Senior
Corporate Counsel" for this teensy company named "Novell."
That got resolved around March, and requires us to pull any 'tainted'
binaries by 7/31. So, theoretically, we could have gotten on with
the release, and kicked it out the door, and done another one once
we got the 4.4-Lite code.
Unfortunately, that was no longer practical, because of other demands
on our time. (for me, it was bad timing, because i was a student,
gearing up for the end of the semester...) We also thought it was
impractical to do a release, then do another two months later,
mostly because the release cycle is several weeks long itself.
So the release was put off until we got the 4.4-Lite tape, integrated
it, and got the system back to "normal." After this was done --
around the end of May -- we could have gotten the release cycle
going again, but i had the problem of having to move across the
country, and for various other reasons, that, too, was a bad time
for people. So we started gearing up for a release at about the
beginning of July -- a full 4 months after we wanted to have
the release out the door.
My goal is to get a release out once every six months; any more
is too much of a drain on the development cycle. Because of the
legal issues that had to be resolved, we more of less had to
skip the 'assigned time slot' for the last release cycle.
Our original goals for 1.0 were along the lines of:
[general] more supported architectures/unified
[general] shared libs
[i386] better device, autoconfiguration, and interrupt
a couple of things that i've forgotten by now, and, of course,
as many bugs fixed as possible.
In reality, "4.4-Lite" integration had to be added to that
list as well, and that's been a _very_ large part of our work
over the last couple of months. We've sent CSRG on the order
of half a meg of diffs to fix bugs in the 4.4-Lite code we've
integrated so far, which is more or less _only_ the kernel
and a few user programs! It's _not_ been easy going.
I actually hope that the next release (1.1) will happen sooner
than six months from now. Some of the goals for it include
complete 4.4-Lite integration, more device support (for all
architectures), more stability, more uniformity across the
architectures, and better install tools. There are obviously
more things that are on the todo list than that, but those
are what pop into my mind immediately.