DISCLAIMER: i was slightly annoyed by this comment. feel free to ignore
the rant below to avoid distraction form important things, like
enlightenment development for example ;-)
On Mon, 2002-05-06 at 20:13, cpg@... wrote:
> unfortunately you don't understand open source. there is plenty of
> empirical evidence that early/often works very well. as you point out
> too early implies some times buggy but there are mountains of evidence
> that that actually entices people to pitch in and projects converge
> much faster.
oooohhh, i would be careful with that statement. it's not too polite to
implicitely categorise someone as being ignorant!
and anyway what's the empirical evidence? i've never seen a truly
controlled study that proves the point that early/often works best in
every situation (emphasis on the "truly controlled"). the 2 succesful
stories in CatB were linux and fetchmail, and only linux is a very good
example since it started from virtually nothing. the basis for fetchmail
was another application, a template for the development of fetchmail.
another *stable* application that is.
a empirical study would take more examples and compare the same
product, either developed in a closed enviroment, or in an open source
context. firstly, to my knowledge, such studies have not been performed,
and secondly they would be very difficult to interpret given the
multitude of variables in the system. the reason you, me, and others
believe in opensource is based on a gut feeling. not on empirical
evidence. it always surprises me how easily people follow other peoples'
arguments and proposed models and then present them to the outside world
as absolute truth. i deal a lot with this in my job as a researcher in
molecular biology, and i can tell you, dogmas are not a good basis for
solid advance. break them down whenever you find a counter-argument, and
try not to present a new one just because you think it's the truth. your
findings are only a closer approximation of the truth, and that's only
when you're lucky.
furthermore, virtually all "open source" projects discriminate between
the early/often version (yes they are called development trees) and the
stable version. gnome does this, kde does this, mozilla does this. just
some random examples of *very* sucessful open sourced products. the fact
that these versions are in the linux kernel tree with odd numbers is
just a reflection of linus' aversion towards cvs >;->. attaching a
number to a certain development stage doesn't mean a thing IMnotsoHO.
but: even when you're completely right about the early/often release
policy it's still a futile argument. remember that E17 is a pet project
of a limited set of very critical developers. they release early (CVS)
and release often (CVS), and when they have a new branch in development,
they even release in parallel (CVS SPLIT & HEAD). the fact that they
don't put any version numbers on it and publish that on say, freshmeat,
doesn't mean that there are no new versions. if you're a talented enough
programmer, you can (oh the joy) even join in.
i'm sorry if my comment sounds offensive to you, if so, please don't
interpret them that way. i see too many people telling others what to
do, and that sometimes bugs me. i think it's better to undertake action:
if you're a coder, why don't you step in and help them release more
often :-) if i was able to code the way the E developers code, i'd jump
in immediately. unfortunately, i can't and don't have the time to learn.
but i don't complain either!
ps. i'm not a contributer, and my opinions might differ from the
developers'. it's just the two eurocents from an interested watcher of