Dave Storrs <[EMAIL PROTECTED]> writes:
> Is anyone here familiar with the behind-the-scenes process and politics
> of the Linux development community?
Not heavily familiar, but I know some details. (My knowledge is that of
someone who's been following linux-kernel sporadically for a year or two
and reads the Kernel Traffic summaries fairly religiously, as well as pays
attention to Alan's diary and various other natterings on the topic
whenever they come up, as I have a strong interest in open source project
> If I understand it correctly (and I'm not sure I have the details
> right), when Linux was being developed, Linus came up with a skeletal OS
> based off of MINIX, then he turned it loose. People volunteered bits of
> code which Linus integrated into the OS. As the thing snowballed, Linus
> couldn't handle the volume, so he chose a group of lieutenants who also
> had commit authority.
Not quite on the last part.
Linux has no version control, believe it or not. (Well, various people
are tacking version control systems on the side, but as I understand it
Linus himself doesn't use version control and all of those other trees are
either for subprojects or after-the-fact imports.) So there isn't such a
thing as "commit authority," quite.
What Linus seems to do more of is just ignore patches unless they come
from the right people or strike him right, and in particular if someone is
listed as the maintainer of a piece of code in the appropriate file, Linus
expects people to send patches for that code to that maintainer and wants
to incorporate patches from that maintainer. But all patches get reviewed
by him, and even Alan Cox gets patches bounced by Linus on a pretty
> Now, I'm not sure where the planning came from...was there a specific
> website or elist or whatever where Linus said "ok, we'd like a TCP/IP
> stack that has the following interface?" Or what?
What I've seen on linux-kernel is that getting new ideas isn't the trouble
at all. *Rejecting* new ideas is most of what Linus has to do. The times
when he sets policy on something like that have all been on linux-kernel
that I've seen.
I would be a bit reluctant to adopt Linux's general development model,
since quite a lot of it as it currently works rests on the fact that Linus
is superhuman. He can maintain quality code without version control, he
can debug kernels without a kernel debugger, he can manage patch
submissions without a bug database, and he can keep track of what he's
doing without a change log. It works for him, so more power to him, but
that really is powers and abilities far beyond those of mortal man, and I
don't think we can expect the Perl maintainers to have similar god-like
> Whatever they did worked, and I haven't heard anything to suggest that
> the Linux community feels excluded...
You've missed a *whole* bunch of political arguments, I'm afraid. It's a
standing conspiracy theory that Linux development is being dominated by
Red Hat to the exclusion of other distributions, etc. etc. Very similar
to some of the ActiveState conspiracy theories that I've heard, actually;
in fact, in some of them, you could do s/Alan Cox/Sarathy/ and get
basically the same rant, complete with the assurances that they consider
the actual tool of the Great Enemy to be a wonderful person.
I think this is par for the course in large, high-profile open source
software projects, regrettably.
Russ Allbery ([EMAIL PROTECTED]) <http://www.eyrie.org/~eagle/>