[Zope-dev] Re: Fishbowl?

Paul Everitt paul@zope.com
Sun, 02 Dec 2001 11:59:18 -0500


Joachim Werner wrote:

> I think that there ARE problems that can not be solved on a mailing list or
> in the fishbowl. One of them is doing a good general design (which we MIGHT
> need for some of the Zope 3.0 issues). I followed all the stuff about the
> CMF and formerly PTK and knew that it was heading to a direction I didn't
> want, but at the same time I felt that it would not help if I just
> contributed to the mailing list. Maybe this was a personal problem of mine,
> but I don't think so.


I don't think so either.  I think your paragraph above does a wonderful 
job of concisely summarizing the challenge.

First, there shouldn't be Annointed Tools.  We should strive to have 
good tools, and we should strive to use good tools, but the real goal is 
communication.  If the current approach isn't hacking it, we need 
something else -- which could mean we learn from successful patterns in 
other projects.

Second, when communication reveals an issue -- what happens?  Let's say 
that every single person in the world of Zope agreed that the CMF was 
going in a wrong direction (just for the sake of argument, as the CMF 
has people that like it as well as dislike it).  Would anything actually 
happen if consensus was reached, and who would be the ones to convert 
conclusion into code?

Third, as Brian pointed out and you conclude with in the paragraph, 
frustrated people tune out.  This causes the other side of the 
communication to get frustrated and stop communicating.  Then things 
break down.  It's important to recognize this is happening, put aside 
the frustration, and address the problems.


> IMHO, there are two possible approaches to problems like that (major design
> issues I mean):
> 
> a) dictatorship, if the dictator is really good in his job (e.g. Jim Fulton
> has done a great job with regard to the design of the ZODB )
> 
> b) meeting in real live (or at least in real time)
> 
> Some of the core architecture of the KDE KParts component model was
> developed on the KDE 2 conference AFAIK. I think we might have to do
> sessions like that at the upcoming Zope/Python conferences ...

That's a very good point.  It's even a good point inside ZC.  Getting 
ten people in a room for an extreme programming session has done wonders 
for our ideas on Zope3.  Anybody want to fly to Virginia? :^)

Yesterday morning I started hanging out on the #zope IRC channel. 
Already it has been illuminating.  It also creates an atmosphere of 
understanding.  I need to do this more often.

--Paul