From: Fred Maxwell <rockbox_at_anti-spam.org>
Date: Wed, 17 Dec 2003 05:58:59 -0500

Daniel,

I'm a professional software developer with over 20 years of experience and I
have to take issue with your philosophy.

> I disagree. I want a large crowd to run daily-builds.
> This way we don't have to deal with beta-releases at
> all. Like for the 2.1 release. This way, we
> get bug reports earlier and thus fix problems earlier.

Bad news for you: There isn't a "large crowd" running the daily builds.
The number of users running a daily build is not statistically significant.
Just ask on here how many people are running the daily build. Do a survey.
In fact, check the server logs to find out how many people downloaded
today's daily build.

> Sure we do. We just ask people, or we all try it for a while, or we find
> the problem and see if that same problem was present already in the 2.1.

Or you get a report of a serious problem and, because it's a very rarely
occurring problem, you don't know if it was in 2.1 or only present in one or
more daily builds. And you and everyone else are left with a vague concern
that the released code has a serious flaw.

> The reason "major software companies" do that, is because they're slow,
> they're closed-source and they have no connection with their users. Also,
> "major software companies" work for money/profit, we don't.

Working for profit means doing things efficiently and intelligently. If
having end-users run daily builds was the answer, then that's what companies
would do.

> We're open, we have a huge crowd of very technical and skilled users and
> we have a huge amount of developers.

Define "huge" in reference to users. How huge a crowd of users is there
running today's daily build? How huge a number of developers are there
working on the core Rockbox code (not including plug-ins)?

> I can't think of any specific problem we've got during Rockbox existance
> that would've been handled a lot better if we were doing what you suggest.
> Can you?

No, but that doesn't mean that I'm wrong. I know skydivers who can't think
of a specific jump which where they needed their spare chute, but it doesn't
mean that jumping without one is a good idea.

Managing beta testing and S/W quality control is an art and a science. It's
based on the software complexity, number of hours of use by the average beta
tester, the quality of the feedback that you get from your beta testers, the
knowledge of the beta testers, and so forth. Simple fact: The largest,
most successful open source projects follow the traditional beta/release
procedure. Look at Mozilla and OpenOffice.org for two great examples. You
don't see those projects skipping beta cycles or encouraging users to run
daily builds. You want to see "huge"? Just look at those projects and that
should give you some perspective on Rockbox.

Regards,
  Fred Maxwell
Received on 2003-12-17