The Simple Economics of Open Source

[Andrew M. Kuchling]

> Robin Becker wrote:
> > I prefer not to worry about the exact electronics/chemistry behind my
> > computer. So in that sense my 'open sourceness' has limits. I guess this

That may be only partly true.  Python is the only free software I've
done serious hacking on.  But I've contributed minor patches to a
number of projects -- mod_java, PCRE, SANE.  They might be only fixes
for NULL pointer derefs or documentation typos, but those other
projects have gained a teensy bit more polish from the fact that I
used them.  Of such incremental improvements is great stability made.

Similarly, people who write software on top of Python, or Zope, or
Linux, can trust that *someone* will take care of the maintenance of
that substrate for them, and, in the unlikely event all the
maintainers go away, you could still do it yourself if needed.

"Gordon McMillan" <gmcm at hypernet.com> writes:
> internet technologies)? Remember when IBM truly believed the 
> world would communicate through SNA? Remember when 
> damn near any real app cost at least $10K?

Also, ever had to debug a problem in a binary-only library through
trial-and-error?  You can burn lots of time that way.  <core dump>
"Hmm... maybe that parameter can't be NULL."  <core dump> "Maybe that
struct field should be 1." <core dump>.  I like the fact that, if I
start seeing mysterious errors, I can dive down into the language
interpreter, farther down into the C library, and even into the kernel
if needed.  Though I've only followed a problem all the way into the
Linux kernel once (to discover it was my error), it was reassuring to
be able to do so.

-- 
A.M. Kuchling			http://starship.python.net/crew/amk/
He spoke to the scar-dancers, to the straw-dust-women, to the old man with a
swan's arm who tends the back stairs, to the three children of the autopsy, to
the painters and the scriveners and the walls.
  -- From a week in Dream's life, in SANDMAN #64: "The Kindly Ones:8"