Re: [Gnu-arch-users] [OT] funding free software R&D

[Tom Lord]

    > From: "Stephen J. Turnbull" <address@hidden>

    >     Tom> At today's prices, RHAT's in particular, I think people _are_
    >     Tom> paying for something that they use, even if they never call
    >     Tom> the support number.

    > But as far as I know much of what you mention (security patches,
    > enterprise infrastructure) is available on the ftp site for no
    > charge.

Not so.  The "bits" may be free -- but the integrated service which
includes: handy client software, contractual relationship to a release
for which you bought support, implied relationship to rhat's
reputation in a commercial context, and relationship to commoditized
IT jobs and certification programs -- is not free.  It's a case of
"pay a guy who knows how to save a few bucks on invoices by
downloading for free" or "create a position whose job description is
defined by a certificate and which, contractually, acommplishes the
same effect (may have to pay a little for some invoices from rhat)."
Smaller shops and larger shops might make that choice differently,
sure.


    >     Tom> It would be weird, then, that, at least arguably, nobody is
    >     Tom> providing a better CVS or even working on one.  A "different
    >     Tom> CVS", perhaps.

    > I think subversion has preempted the "better CVS".

In my perception, there's a growing disconnect between the
historically advertised goals of svn, and what counts as "things to do
before 1.0".    I suspect this relates to the people paying for
development having a narrowly specific set of applications in mind.

When I say "a different, not better, CVS" I'm mostly referring to the
trade-off of features and admin costs.  A common gripe about CVS is
the admin hassles of its vestigial flakiness -- I don't see svn
obviously making that better and arguably making it worse.  Another
common complaint is security issues -- and I have the same response
(simply based on LOCs-of-much-newer-code if nothing else).  Another
common complaint about CVS is performance and here I have a similar
observation: different performance trade-offs rather than clear
across-the-board improvements -- but the qualification that in most
areas of performance they have the greatest chance of getting their
act together before 1.0.  (The qualification to my qualification about
performance is that the rate of Berkeley DB log-file file growth is a
nasty space-performance / I/O performance / robustness trade-off that
I don't see them having a great chance of doing anything about prior
to 1.0.)

When I say "a different, not better, CVS" I'm also referring to the
postponement of smart-merging features and the lack of clarity in
their stated plans for those, post 1.0.


    >     Tom> I think it's because there's not much demand for software
    >     Tom> development and, consequently, tools that help make software
    >     Tom> development _better_ are anthema.

    > There's plenty of demand for software development.  On the one side,
    > there are the big proprietary firms and the inhouse IT organizations.
    > But for them a few hundred seats of Rational license is a
    > non-issue.

I suppose "plenty" is your weasel-word there.   Have you _seen_ the
unemployment figures?  The revenue figures?   I wouldn't expect you to
but I suppose you also aren't thinking about the big cultural changes,
_not_ limited to but certainly acute in silicon valley where, not all
that long ago, employers were fostering new software projects as fast
as they could get them.


    >     Tom> And please note that "commercial demand" and "what hackers in
    >     Tom> the free software community talk about wanting" are very
    >     Tom> different things.

    > Of course.  But ...

    >     Tom> Evidence, as good as any, of the _failings_ of RHAT execs to
    >     Tom> align their business model with the available resources.

    > Please unpack.  I don't get it.

Same old, same old.    Some software architectures give hackers more
inspiration to participate and generate value than others.


    >     Tom> The rule of thumb for the tech industry is "something
    >     Tom> surprising will happen".  I think that rule of thumb is a bit
    >     Tom> like Moore's Law: it wears out -- in our lifetimes -- soon --

    > What makes you so certain either is going to wear out?  You sound like
    > the late 19th century physicists who declared that everything that was
    > left was "engineering" (although Michaelson-Morley etc had already
    > been done).  Yes, I know about the "absolute" physical limits, but
    > they assume things like von Neumann architecture, no?

Nothing makes me certain.  What I was specifically thinking, re
software, is simply that the number of lines of new or newly
integrated code needed to make something that will have large impact
seems to be growing very rapidly.   This can be contrasted with 20
years ago when a few 10K lines could change the world.

That makes intuitive sense to me because: 

(a) the expressive power of our programming environments isn't growing
    very rapidly

(b) the environment in which programs have to operate to succeed is 
    growing very rapidly (e.g., it's not a stdin/stdout/stderr world,
    for the most part, to state it in exaggerated terms)

(b) if you hold expressive power constant then it doesn't take too
    long to cherry-pick the space of programs that an individual or
    small team can make;  if you complicate the environment, that
    accelerates the process

Once again, at least in the domain of interactive applications -- the
emacs architecture is the way forward.  Yes, it will take a lot of
work to make a new one with enough performance and eye-candy, and
graphics features to have impact -- but once it's there it's a tool
that _does_ (greatly) extend the expressive power of our programming
environments, and therefore creates a new grove of low-hanging fruit
(little hacks that can have big impact).



    > I think there are plenty of surprises left in our lifetimes.

Perhaps.  A lot will depend on politics well outside of the tech
industry.   Will we wind up in Terry Gilliam's Brazil or Rick Burman's
(pre "Enterprise") Star Trek?


-t