All over the place. Most developers work on it in their spare time - but only members of the core team have commit privileges - which is the right to make changes to BSD. This is a fundamentally different approach to the widespread anarchy of Linux :).
> When do they find time to develop software? Similarly for the Linux
> kernel, GNOME, or Mozilla.
Most developers work in their free time. Only a handful of companies hire people to work on core Linux. SGI has some people working on the port of Linux to MIPS, although many of the Linux on MIPS team are "mere" enthusiasts. VA Research (a hardware vendor) has hired some core developers. Transmeta (a secretive firm doing R&D in microprocessors) has hired some core developers, including Linus Torvalds himself. Red Hat has hired a few.
For that matter, for a brief shining moment I was a person paid to develop a tiny bit of core Linux - I made some changes to the intel binary compatibility system. Needless to say, this is quite rare (both being nuts enough to tackle iBCS and lucky enough to get paid for it) and only lasted long enough for me to get a SCO binary to run properly.
Mozilla is something entirely different different; most of the Mozilla core team is employed by Netscape/AOL. This is definitely worth noting.
GNOME is almost entirely enthusiasts; the GNOME leader is Miguel de Icaza, whom I've mentioned before, who works on it in his spare time. A couple of the core GNOME people work at Red Hat in RTP.
> In my experience, one motor for development is the fact that systems
> people write their own software, and then share it. That's how Perl (the
> most popular language for WWW site programming) was developed. This
> doesn't, however, explain everything - systems larger and more complex
> than Perl seem to use a more co-ordinated effort. E.g. free operating
I don't think so, really. Firstly, perl is an incredibly complicated language, and with the plethora of modules and libraries much more so. Larry Wall wrote Perl in 1986 in concert with several other people at the JPL. Still, that's the traditional core group model, merely informal.
Really, the heroic image of the individual hacker contributing software to the world is very much a thing of the past - the first few versions of gcc or GNU Emacs, Richard Stallman's master opus, were like that. (You could also count Wall's patch utility here) However, it's impossible for that to be kept up; Stallman himself hasn't done much active development since he developed RSI in the early nineties.
> My feeling is that to some extent this 'way of organising' sometimes
> involves resistance by computer professionals against the pressures of
> capitalism. I know that as a systems programmer, I prefer working with
You know, while there is a strong sense of resistance to capitalism, I wouldn't call it a leftist streak. For the most part, it's a strong right-libertarian streak, as exemplified by Eric S. Raymond. They don't so much reject capitalism as reject restrictions on *their* rights and *their* priviledges - it's a wholly atomistic worldview.
(It is ironic that Torvalds and Stallman are counterexamples to this philosophy, but they are very much politically isolated in this way).
> maintain control of the computers. By contrast, large software purchases
> normally go hand in hand with the involvement of outside consultants, and
> lessened ability for in-house computer people to control the process of
> production.
Less so than you'd think. While the ability for in-house people to control the process and means of production is there, it's minimal, because the expertise and the time isn't there. It is a step forward, but it's still a small one so long as programmers are worked to the bone and have no opportunity to seize the means of production. This is the great flaw of the GPL - the opportunity doesn't equal the ability to use the source.
Consider my present experience. I work at a consulting company, doing database-backed websites. We usually distribute our work with code exposed (rather than concealed - Notes lets you do that); our clients can make whatever changes they want. However, it's exceedingly rare for them to do so. They don't have the experience and they don't have the time.
> open source software. Richard Stallman's vision with the Free Software
> Foundation was the create a situation where software is liberated from the
> bounds of money. Raymond's vision of 'open source' is more along the lines
I don't think that that's precisely so, even though that's the way it's commonly interpreted. Stallman has never been above accepting quite exorbitant - and I do mean incredibly exorbitant - fees for his time, porting gcc to esoteric architectures. What the GPL does is not divorce software from money, but the control of the software and the software's future always lies in the hands of its posessor rather than in the hands of the license owner.
Consider that in the early eighties, when Stallman was a doctoral candidate at the MIT AI lab, his fellow students and coworkers were being hired hand over fist by commercial ventures, and their work lost. The FSF was his way of rejecting that, and the GPL the means to reject that. In a land where the GPL applied to all software, you would have to earn your keep by pushing science and technology forwards rather than rewriting the same thing time and time again.
marco
,--------------------------------------------------------------------------.
> Marco Anglesio | There is no premature anti-fascism. <
> mpa at the-wire.com | --Ernest Hemingway <
> http://www.the-wire.com/~mpa | <
`--------------------------------------------------------------------------'