Hi!
Dan Christopherson wrote:
> On Sun, 4 Feb 2001, Rickard [iso-8859-1] Öberg wrote:
>
> > Jason Thomas wrote:
> > > Then again maybe some of that 95% quit trying to help because their
> > > ideas are made fun of.
> >
> > I said that the design of the system was most excellent, just not
> > appropriate in this context. Anything else is your own interpretation.
> Your exact quote was "Excellent design of a silly control system"
> How was Jason supposed to interpret the word 'silly', other than as a
> flame? I'm not saying that it was neccesarily your intent to come off this
> way, but it's probably how most of us would have interpreted your
> statement (It's how I interpreted it, at any rate).
Yes, that was an unfortunate slip. It was not intentional. Whenever
possible I try to keep an open mind about the path's that are available;
it's just that the proposed system is about as far away from what I want
that is possible.
> That's not to say that the proposed system didn't fill me with a certain
> feeling of distaste: the notion of 'designers' throwing 'detailed design
> documents' over the wall in an open source project gave me the willies: I
> like open source because that doesn't happen, because it's assumed the an
> individual programmer will perform detailed design. It's true that someone
> more experienced might encourage modifications to this design, but that's
> what this mailing list is for. This also allows all parties involved
> (including lurkers) to learn constantly. IMHO, if you don't have a desire
> to constantly learn, you shouldn't be in this field.
Agree completely.
> 1. Find a bug
> 2. (Optional) consult with jboss-dev
> 3. Kill it
> 4. Submit your patch
> 5. Find out what was wrong with your patch
> 6. Repeat 2 through 5 as neccesary
You could have 3 before 2, and 3 as a fork when not able to squash it.
Depends also on how easy it is to kill, and if it might have
side-effects that has to be considrerd. Tricky business.
> 1) Step 2 will be much easier once Bugzilla is back (Or a Bugzilla
> replacement 'Powered by jBoss'?) This should also allow developers to
> check out bugs or features for themselves (I don't think there
> should be any 'board assigned' tasks - Telkel employees might be
> assigned tasks, but that's different). This might be a big change: how
> many people consult bugzilla before reporting bugs? I tend not to, but
> that's might just be me and my feelings about bugzilla.
Personally I think the current BugZilla is (unfortunately) not working,
from a development process point of view. It needs to be simplified, and
get a better UI. I don't know when Scarab (scarab.tigris.org) will be
available, and I'm not at all sure I like its architecture (very
non-J2EE-ish AFAICT) but I know that Simone hinted that he wanted to do
something similar.
> 2) Steps 2,4, and 5 will really only work when module/subsystem owners
> commit to replying to messages within a reasonable timeframe. This doesn't
> mean that other people won't discuss it (by now you've realized you'll
> never shut me up 8^})) - it just seems to be needed to ensure response.
Yes, true.
> 3) Note that I don't assume that every contributer is a commiter (cvs
> wise) My personal opinion is that this probably isn't neccesary, that
> making people submit patches has a couple of benefits (in the 'more eyes'
> territory) and that anybody who really should have commit priviledge will
> get it once they've buried a subsystem owner in patches 8^})
Yes, definitely. Getting eyes to see if the patch might have unwanted
side-effects is crucial in such a large system as JBoss. The modularity
definitely helps in minimizing these effects, but still.
> Well, that would be my .02$ (and you got a lot of text per penny, if not
> much else)
Yes, that was a very valuable .02$ if you ask me ;-)
thanks and regards,
Rickard
--
Rickard Öberg
Email: [EMAIL PROTECTED]
