[Mailman-Developers] ZMailman 2.1 preview

[Stephan Richter]

At 06:50 PM 12/9/2001 -0800, Dan Mick wrote:
>JC and Stephan's points describe differences in internal implementation
>and organizational development, but I'm having a hard time seeing actual
>advantages there.   Security is the one thing I see as "better with Zope".
>Surely there must be other compelling reasons; I mean, one could recode
>Mailman in Perl, but why?...one could change from native FS to Andrew FS,
>but why?...etc.

This is a good point, however I think the current mailman data structure 
looks like a big hack (I am sorry to say that, but you can clearly see that 
Mailman grew over the last years). So they need to be reimplemented again 
in anyway and I would have done that already, if it would not be such a 
huge task, but I hope to be able to do it soon with some help... maybe I 
get something prototypish done over Christmas, if not definitely something 
will happen after that...

Having said that, I think that the data structures should be savable in any 
data source, be it an RDB, FS, ZODB or BK. I just think that natively (or 
the default) the ZODB would be a much better fit with Mailman, since they 
are both developed in Python. Please remember, I do not talk of Zope, but 
*only* about the ZODB.

Here is one of my main arguments: Many people of the Python scene develop 
the ZODB (including Barry) and they solve many storage problems for us. So 
why would should Mailman worry about its own storage mechanism and put 
development resources on it, if we (Mailman Development Team) could 
"out-source" these expertise?

BTW, calling Lock(), UnLock() and Save() all the time I made a change to a 
data structure in Mailman seems very clumsy to me, since I am very spoiled 
by the ZODB behavior.

Regards,
Stephan

--
Stephan Richter
CBU - Physics and Chemistry Student
Web2k - Web Design/Development & Technical Project Management