| Progress on Dynamic Content Manager|
Over the weekend, I ported most of the classes and rules from the Kamat Content CrowBot to its Web version. I realized a grave mistake I had made early on by relying
a story's location to determine its relevance. This was OK
in the old days because an artile on paintings was always
under a folder called art. But with the dynamic content management, there are no physical files
The biggest problem was that over the last two years I have written so much SQL of the kind Select * from Contents where URL like '/art/' to build the table of contents, and with dynamic management, I have to rewrite those queries!
Dejected, I consulted Hiryoung as she prepared chiken curry. "Why not implement virtual folders?" she suggested. I could, except that there would be 250 (the current number of folders on Kamat's Potpourri) of them, and then I'd have to write a resolver...
So we decided to take an approach that would be very specific to us (hear me loud ladies and gentlemen -- custom content management), by aliasing the dynamically generated URLs to the traditional slash (/) seperated, internal representations. This way, my legacy SQL can stay as it is, I do not have to rewrite my CCE (content classification engine), at the same time start deploying the dynamic CMS.
I also implemented a multiple pass feature for the content management. The first pass sets some state that will be processed by the second pass. It is like the software that writes software. Very cool. Works very well... works very slow...
|(Comments Disabled for Now. Sorry!)||First Written: Sunday, August 26, 2001|
Last Modified: 1/29/2003