Date: Thu, 4 Nov 2004 11:38:04 -0800 (PST)
Message-Id: <200411041938.iA4Jc4UX055425@xl2.seyza.com>
From: Thomas Lord <lord@emf.net>
To: stephen@xemacs.org
In-reply-to: <87y8hinbq1.fsf@tleepslib.sk.tsukuba.ac.jp> (stephen@xemacs.org)
Subject: [OT] Re: [Gnu-arch-users] Re: Re: community spirit
References: <200411010336.iA13af6p023128@xl2.seyza.com>
	<20041101141701.GB9161@puritan.pcp.ath.cx>
	<87wtx51w70.fsf@beeblebrox.rfc1149.net>
	<20041101144826.GD9161@puritan.pcp.ath.cx> <41867B20.1020608@melon.dk>
	<2982EA60-2C38-11D9-8C6F-000393CFE6B8@spy.net>
	<4186917E.3030209@melon.dk> <20041101232959.GA25080@hezmatt.org>
	<87ekjcswdn.fsf@tleepslib.sk.tsukuba.ac.jp> <418765A9.2040206@diku.dk>
	<878y9kpfcm.fsf@tleepslib.sk.tsukuba.ac.jp>
	<200411022006.iA2K6Q1n036262@xl2.seyza.com>
	<87y8hinbq1.fsf@tleepslib.sk.tsukuba.ac.jp>
Cc: gnu-arch-users@gnu.org
Precedence: list
Sender: gnu-arch-users-bounces+migo=homemail.com@gnu.org
Errors-To: gnu-arch-users-bounces+migo=homemail.com@gnu.org
Status: RO
Content-Length: 5840
Lines: 114

    > From: "Stephen J. Turnbull" <stephen@xemacs.org>

    [Reuse: good v. bad]

I think we mostly agree but for word choice and emphasis.  I get too
many lamebrained suggestions from people that problem X can be solved
by combining components W, Y, and Z where those components themselves
have 10 dependencies and so forth ---- the danger of non-critical
reuse is accute among programmer's who don't insist on having ready
access to a detailed understanding of how the resulting system they
construct works.

    > And it's not just you.  There are dozens of smart free software
    > people who know a thing or three about system building, some
    > with enormous entrepreneurial talent.  Take the leading example:
    > rms.[1] But then, you just deprecated GNU out the wazoo, didn't
    > you, so I guess he, too, whiffed on the bloat-ball.  On the
    > other hand, none of the *BSDs offer nearly the range of software
    > that GNU/Linux does---unless they offer ports of GNU software.
    > In fact, by now all of them have rather accurate Linux
    > emulators!

The open question is whether a large number of users' feature
desires/requirements could yet be or could have been solved by taking
some /other/ approach than the chaotic, out-of-control bloatware we
see today.  Could there be a desktop that is as slight and crisp as a
traditional BSD core system distribution?   Or is there some actual
domain-specific need to have desktop systems which are basically out
of control?  (Or perhaps we need a private MSFT-scale campus if we
want to compete in desktops?)

The bloated crazy approach is promoted in the industry press in the
form of which projects get the most attention and in terms of what
kind of attention they get (and, as a matter of history, that press
attention does /not/ follow consumer uptake -- it precedes it).  That
industry press is in turn controlled by members of the same social
clique of execs who are checksigning for the crazy approach:  Does it
make me paranoid to think that the press and those execs share a
common reality-perception bug wrt. how software works?

Meanwhile, since the time I was a young whippersnapper -- a junior
programmer barely competent to login -- i've heard tales of and later
experienced the forcible supression of efforts to take a different
path, usually that supression being justified by vague,
pseudo-economic arguments about what (the speaker wishes we'll
believe) the various relevent markets demand.  One very famous
old-school programmer said to me once, slight paraphrase, "I've never
managed to fix [those execs] -- I doubt you will either."  He seemed
to me to have slipped, very largely, into a hopeless cynicism.

Well, "second generation" is a pretty interesting position compared to
"nth generation".   When first generation says "I'm not sure you can
fix X", where X is obviously of central importance, to the second
generation: that's nothing but a priority challenge.


    > Sure, we all know how to do it in theory, and given the resources,
    > people like you surely could implement it.[2]  But the fact remains:
    > nobody has managed to do it.  Nor does it look like you're going to
    > anything effective about trying to do it yourself---you've got better
    > things to do, like Arch, and Pika Scheme, and xl1, and furth, and so
    > on.  *And everybody else is just like you*: they have more important
    > things to do (and of course they want them packaged for the distros!)

    > No?

Stay tuned.  My multi-media document format is shaping up nicely,
recently (at least the infrastructure and structure for building it
out).  And that's being done as part of the Arch project which it will
begin to benefit shortly and amply.

Such frugal and serendipitous parsimony is great (and lucky) when it
works ("ain't i clever?") but it is a sentimental fantasy to believe
that such cleverness should compete, alone, against the market and
social forces constructed by the current checksigners.  Such
cleverness is, by and large, a crapshoot (although some of us are
pretty good at spinning the dice if we're only rolling a few times).
To the degree that /my/ design approaches can be said to represent
those of a larger community and/or a specific and plausible design
mentality -- my team has been consistent very, very badly outspent and
out-promoted, often disrupted, often ripped off, often treated quite
rudely: we never stood a fair chance to compete.  There is, in other
words, no level playing field of software design ideas from which
wanna-be software execs can choose and evaluate potential investments
on the basis "available packages for binary download".  There is /no/
way for those execs to behave rationally in their check-signing
ability unless they are able to dig, pretty darn far, into actual code
and code designs.  Too few of them are qualified, let alone
interested, in such digging-in, as far as i have been able to observe.

Where does that leave such willfully ignorant execs in my moral
calculus?  Well, how would you expect me to feel about people who are
doing engineering, installing networked machines in 100s or thousands
or millions of sites around the world, using propoganda to shape
market perceptions and expectations, yet taking so little visible care
about the quality or environmental impact of those machines they are
deploying?  Where I come from, we call that kind of engineer a "tool"
and we do everything we can to wake them up or else regretfully beat
them back with a stick.  Even if that "everything" is sometimes pretty
humble and gentle like a flame on g-a-u.

-t


_______________________________________________
Gnu-arch-users mailing list
Gnu-arch-users@gnu.org
http://lists.gnu.org/mailman/listinfo/gnu-arch-users

GNU arch home page:
http://savannah.gnu.org/projects/gnu-arch/