[tex-live] free software, DFSG

[David Kastrup]

plaice at cse.unsw.edu.au (John Plaice) writes:

> David,
> 
> Just watch the phoenix rise out of the ashes.

I'd be content to witness a Zombie crawling from a graveyard.  John,
with all due respect, you misestimate the problems of Omega.  If you
want to get a clue about what problems Omega2 is up to, you need to
look no further than Omega1.

When Omega1 came out, it tackled several problems for typesetting
that had been ailing TeX.  Why didn't it take on?

a) LaTeX support.  LaTeX offers its own ways of dealing with input
encodings.  One would need to come up with some conventions or
packages for dealing with LaTeX-related input.  This entails talking
with developers and having programmer-accessible documentation.

I am not the most stupid person on Earth, and I tried for some days
at that time to get a hang of OTP, sufficient for getting a clue how
one would integrate them with LaTeX, and failed.

b) output driver support.  One needed special programs oxdvi, odvips
and so on to deal with Omega.  The respective changes should have
been submitted to and folded back into the main drivers.  From time
to time, compliance with Omega should have been rechecked.

c) user support.  Users need instructions and example files to get
systems working.  My attempts to find such instructions, or even get
volunteers accepted for _creating_ such stuff were futile.

d) accountability.  Whenever somebody would have dared saying
something like "I want to base a major system off Omega.  If I
encounter problems, what do I do?".  With Knuth, the answer is that he
will pay a finder's fee for any bug.  The maintanance intervals
nowadays have reached a proportion where it is clear that the effort
is fading out, but it was there initially.  With Omega, the answer was
"Don't bother us with that.  With the next big rising from the ashes,
it will all be obsolete, anyhow."  So we do not even have a proper
multilingual LaTeX format integrated with Omega.

All of these problems will be _exactly_ the same when (and if ever)
Omega2 will magically appear in fully functional glory out of the
void.

Maybe I am factually mistaken about some of the above points.  But if
I am, there is nothing on record to _show_ me wrong.  Even if some of
my impressions might possibly be wrong here, there is nothing I can
find published or in common knowledge to dispell them.  And in the
end, this is what counts.

I am maintainer of several TeX-related projects.  I am a lousy
maintainer, and a lousy worker.  I have moved AUCTeX (a project more
or less abandoned by Per Abrahamsen) to a much more visible place,
have got it accepted as a GNU project, have garnered visibility.  I am
blowing all my self-inflicted release deadlines, but the thing still
moves on.  People have come onboard that finish what I considered my
duties but didn't.  Stuff gets implemented where I am just standing
by.

Take a look at Linus Torvalds: how much actual programming is he
actually doing these days?  His time and expertise is too valuable for
that.  He is a full-time manager mostly, instead, and only actually
touches code when it is of vital importance.

A user and developer community does not magically appear with the
release of a package.  You have to prepare for it and work for it.

With AUCTeX, I have perhaps a user base of 10000 people or more, and
maybe 4 actively visible developers.  I am working on increasing those
numbers while balancing the development process.

With Omega, you have maybe 4 actively visible users (Yannis, Idris,
and who else?).  Regardless of whether they are using Omega1 or
Omega2, this will get you nowhere.  You need to cooperate with driver
authors, library maintainers and so on to make sure that switching to
Omega becomes a well-understood step and process.  And something that
will cause minimal pain when Omega2 will appear.  You can easily make
sure that by the time it appears, its requirements will be met by
xdvik, dvips, and other tools.  And, more importantly, you can make
sure that people _know_ this.  By having alpha testing, by gathering
experience, by trying to have people adapt it to macro packages and so
on.

> And, as you say, if it is not done with Omega,
> it will done by another project.

Which would probably be a pity.  But I feel incapable of checking
that.

-- 
David Kastrup, Kriemhildstr. 15, 44793 Bochum