incubator-general mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Stefano Mazzocchi <stef...@apache.org>
Subject Re: proposal: modify incubation application process to require a reference to the code itself
Date Mon, 12 Jul 2004 14:29:34 GMT
Rodent of Unusual Size wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> 
> Brian Behlendorf wrote:
> 
>>It seems to me that any honest assessment of the merit of
>>accepting a proposal should include a look at the code itself, if only to
>>get a gut-check on how maintainable and evolveable that codebase might be
>>going forward.
> 
> 
> why?  if the idea excites people but the code sucks, are we going to
> turn it down?

I agree with Ken. In my experience, good ideas and bad code work a lot 
better in creating a community than good ideas with good code. And the 
reason is that it makes it a lot easier to suck people in, because they 
like the concept but not the code and start fixing it.

I also agree with Ken that a proposal should just indicate that a 
working codebase is in place and the mentor should be able to attest 
that (so he/she should be given a copy of the software, with source, to 
try it out), but this doesn't require things to be thrown in the wild 
(practice that would be hard for those companies that are willing to 
open source some of their software).

So, in short: I personally think that we *must* require a working 
codebase (incubating communities out of simple ideas is just asking for 
trouble), but that the incubator is not equipped to judge the value of a 
software quality at the acceptance stage, but only at the exiting the 
process.

Basically, the whole idea of the incubator is that we are using 
darwinian principles to evaluate quality and health of a community, 
assuming that without value, such a community wouldn't form around it.

What the incubator "acceptance process" should try to filter is those 
proposals who cannot even make it inside our petri dishes, and code, 
besides being somewhat operational, should not be part of the judgment, 
or we make the mistake of doing some of that darwinian selection ourselves.

And why would that be a mistake? in my experience, projects could change 
shape when they are incubated (even in the pre-incubation days), one guy 
could come along and just reshape the thing in such a way that at that 
point the community can florishes.

Code quality doesn't mean anything for incubation: it comes up by itself 
later, when the incubation is successful. Very counterintuitive, I 
agree, but that's what I've seen happening far too many times for being 
a statistical singularity.

-- 
Stefano.


Mime
View raw message