Zope (and a bit MMBase)

I've been playing around with Zope (www.zope.org), mainly to get a
feel of how a Web publishing tool works in practice, to see how our
work could fit in and to see how it compares to MMBase
(www.mmbase.org, the tool the VPRO uses for its websites).

I've started a little site at http://www.ossenbruggen.com/ and managed
to change and upload data to the site from my PC at home through
Zope's Web interface (that is, without logging in at CWI.  So, at
least in theory, my family can now help to extend the site without the
needing an account at CWI.  But knowing my family, I'm affraid it will
remain just theory ... :-).  Facilities like this could be used to
have users and/or domain experts add/update media items and metadata
to our hypermedia retrieval/presentation generation system.

Second thing I I wanted to find out was how easy it is to integrate
some existing software into Zope.  A bit of a steap learning curve -
Zope documentation is still "work in progress".  But all in all, it
was pretty easy, and required only a minimal amount of "glue" code.

First results at:

   http://spinaker.cwi.nl:8080/test/

Try to fill a name from one of the people of our INS2 group at CWI
(e.g.  Hardman, Rutledge, Ossenbruggen should all work, Niessink,
etc. doesn't - sorry :-).

The resulting publication page is generated by an "external method" of
the Zope server running on my machine.  The external method is the
glue, a Python function calling the "Bridge" (also Python) software
Frankn and I developed at the VU to manage our bibliographies.  The
software parses our group's publictions encoded in XML and converts
all entries that (exactly) match the query to HTML.

It takes a while, but that's mainly because it re-parses the XML file
each time you try it.  This is in not a limitation of Zope, but of
Bridge (we designed it as a batch conversion tool, not as a long
running process that can input a file once and produce output
interactively).

Also note that I'm using Zope here merely as an interface to Bridge,
where all the interesting stuff happens (XML processing and
searching).  I did not look into the built-in XML and search
facilities of Zope yet, but it seems to have at least some.  This
should improve now Python 2.0 is Unicode compatible and has an XML
package in the standard distribution.

All in all I think I'm starting to like Zope (it installs out of the
box!), except for the fact that - behind the scenes - it stores
everything in it's own database format.  This might make it difficult
to switch to other tools later.  Compare this with VPRO's MMBase,
which stores its objects in any third party SQL database, including
MySQL.

Also note that this is all pretty Python oriented, which is nice in
the case above because the Bridge software is also in Python.  MMBase
uses Java, which might be more convenient in other cases.  Zope's
"DTML" language (sort of HTML templates + Zope scripting) seems to
look just like MMBase's SCAN.  Zope and MMBase feature both
customizable "editors", these are web forms that help you to build the
web-pages.  Zope's security model seems a bit more sophisticated than
MMBase's, with users and roles that can be assigned, inherited and
overridden at every level in the hierarchy.

This is it so far,

Jacco