ZServer design ideas: gneeral comments
Hey all! I recently returned from the IEEE Metadata '99 meeting, and had some thoughts that I nned to share with the Zope developers. First off, I did a little evangelism, recomending scripting in general and Zope/Python in particular where ever I could :-) One particular case involves a software/data archive at a NASA. Under the title "Smart Data, Dump Archive", they've come up wit some thing they call 'buckets' that serve to contain data and associated software to act on that data, that can be moved around between archives, and is accessed through the web. Hmm, sounds like WWW enabled objects with methods, doesn't it? Where have I heard this before? They are currently hand crafting perl/CGI to do all this, so I recommended a look-see at Zope. The one piece they're looking for that Zope doesn't do well right now is the packaging everything up and moving it around part. Mostly because of the 'fear of pickles' issue - err, I mean 'fear of arbitrary foreign code'. A second point from the meeting was about interoperability. The general concensus was "Standards, Standards, Standards, oh and then some more Standards". Admitedly, there's more standards than you can shake a stick at covering all aspects of something like this, but the one application I though of was ZServer. My understanding is that it is ZServer's job to provide different "views" of the ZopeBase (err, what was the agreed on term for this?). One of design issues around the FTP view is what to do with attributes and methods of docs. One suggestion was an XML doc, containing the extra info. All that 'extra info' is what the whole meeting was about, namely Metadata! There exists now a proposal to the OMG (Object Managent Group - the CORBA people) to fold aspects of UML into XML and call it XMI - XML Metadata Interchange. It's a little verbose (what commitee standard isn't?) but there are working implementations in use in the database world, allowing exchange of object schemas and such between different vendors tools. Take a look at: http://www.oasis-open.org/cover/xmi.html an archived version of the announcement of the modified proposal: http://www.omg.org/archives/orbos/msg00702.html To quote IBM (one of the submittees) on the proposal: "The main purpose of XMI is to enable easy interchange of metadata between modeling tools (based on the OMG UML) and between tools and metadata repositories (OMG MOF based) in distributed heterogeneous environments. XMI integrates three key industry standards: 1) XML - eXtensible Markup Language, a W3C standard; 2) UML - Unified Modeling Language, an OMG modeling standard; 3) MOF - Meta Object Facility and OMG modeling and metadata repository standard."
From Zope's point of view, it provides an standard DTD creating an XML representation of the ZopeBase.
HTH, Ross -- Ross J. Reedstrom, Ph.D., <reedstrm@rice.edu> NSBRI Research Scientist/Programmer Computer and Information Technology Institute Rice University, 6100 S. Main St., Houston, TX 77005
Boy, sure can tell I hadn't eaten lunch yet when writing this.
Re: [Zope-dev] ZServer design ideas: gneeral comments general
Ross J. Reedstrom wrote:
Hey all! I recently returned from the IEEE Metadata '99 meeting, and had some thoughts that I nned to share with the Zope developers. First off, I did
need
a little evangelism, recomending scripting in general and Zope/Python in particular where ever I could :-) One particular case involves a software/data archive at a NASA. Under the title "Smart Data, Dump
oh, a bad one: that should be "Dumb Archive"
Archive", they've come up wit some thing they call 'buckets' that serve with something
<snipped remainder, with other spelling mistakes, but at least understandable in context> Ross -- Ross J. Reedstrom, Ph.D., <reedstrm@rice.edu> NSBRI Research Scientist/Programmer Computer and Information Technology Institute Rice University, 6100 S. Main St., Houston, TX 77005
Ross J. Reedstrom wrote: Another theme from metadata '99 was convergence: all the different communities are starting to do the same thing, but coming from different directions. This leads to part of the "The nice thing about standards is that there are so many of them to choose from" problem. In the context of my previous mail, the converging communities are the database people (particularly CORBA object database people) and the WWW community. And their competing standards are: XMI (XML Metadata Interchange) from OMG and RDF (Resource Description Format) from W3C Looks to me like they cover pretty much the same turf, with different emphases. Hopefully, if ZServer is a pluggable and flexible as I think it is, supporting both (eventually) should be an option. The choice of what to implement first may depend on what tools someone needs to make Zope compatible with: db/CASE stuff, for XMI, or the equivalent web site authoring tools, for RDF. Of course, the willingness to spend the time or money to do it or have DC do it will be the final deciding factor. I think I'd bet that an Python RDF parser pops up first. That first category might be more directly applicable to Zope than the average webslinger might imagine, especially since Paul intoned on Friday:
Zope2 will go a long way towards making object storage more confidence-inspiring through (a) change in format and (b) allowing storage in "safe" managers such as an RDBMS or bsddb.
Perhaps CORBA? Zope as a CORBA client, turning objects into HTML? I may be talking through my hat here - what I don't know about CORBA could fill libraries. Hmm, there also seems to be a dearth of Free CASE tools. Perhaps the XML/RDF stuff will fix that, via the back door :-) Ross P.S. I just discovered a benefit of having a conversation with yourself via an email list: I can reply to the list without editing myself out - my MUA does it for me! -- Ross J. Reedstrom, Ph.D., <reedstrm@rice.edu> NSBRI Research Scientist/Programmer Computer and Information Technology Institute Rice University, 6100 S. Main St., Houston, TX 77005
On Mon, Apr 12, 1999 at 04:32:48PM -0500, Ross J. Reedstrom wrote:
Perhaps CORBA? Zope as a CORBA client, turning objects into HTML? I may be talking through my hat here - what I don't know about CORBA could fill libraries.
Actually, this should be quite possible, assuming you do an ORB bridge between CORBA and the interface that Zope expects. How USEFUL it would be is a totally seperat question :-) CORBA is (in most implementations) awefully heavy, and quite ugly. :-) What I see CORBA being good for is as a way to interface a Zope system to nefarious third-party stuff that hasn't been shown the true light yet :-) But this should be hit with EVERY web hit. :-)
Hmm, there also seems to be a dearth of Free CASE tools. Perhaps the XML/RDF stuff will fix that, via the back door :-)
Um, probably because there's a dirth of good CASE tools in general. Lots of crap, very little useful stuff, and that which IS useful is terribly expensive. This is just my two cents from trying to use what's out there. When I worked for LMSC we invested literally millions into this stuff and then flushed it because it was too ridged. UML helps some of this, but it's still too difficult to model lots of concepts. Chris -- | Christopher Petrilli ``Television is bubble-gum for | petrilli@amber.org the mind.''-Frank Lloyd Wright
participants (2)
-
Christopher Petrilli -
Ross J. Reedstrom