Interject away. I read your post with great interest. And, please, you all ;-> were here first, and lots of people call me Charles anyway. Nuno's just used to talking to you.
The code I originally inherited (before I created the query engine) was a grab bag of Perl scripts that used regex to extract tags and values from a flat file (one entry per line). This worked fairly well, but was tricky on the update side and was limited in scale. But for less than 10,000 documents, it was quick, easy and very flexible.
It seems reminiscent of your "MV engine". We did many experiments and made many improvements in our journey towards a query language. I learned a lot about tuning regex in the process. In the end, the flat file was the biggest limitation. We had >100K docs. We also did experiments with SQL and LDAP (which wasn't too bad).
I added support for hierarchical names (going left-to-right, like Java package names) to my own, C++ version. But, from the POV of the query engine, this support is simple minded. It is just "starts with" plus a delimiter. Efficient in both SQL and regex. Again, the update side is trickier (managing the tree, selecting reference values from the tree), but not bad.
take it easy,
Charles
At 07:25 PM 2/9/2003 -0800, Charley Bay wrote:
Charles Reitzel wrote: > Hi Nuno, > > Nuno Lopes wrote: > >Hi Charley, <snip> > > Me and Charley Bay are two different people. I > actually go by Charlie most places. I decided to > go by Charles on this list to avoid confusion with > you, Charley! No biggie.Ooops. Sorry to interject. ;-) I'll try to start going by "charleyb" then. ;-) To further complicate the issue, after much therapy we've found out that Charley Bay is himself several different people. But, they have all agreed to use the one screen name, "charley" (now "charleyb"). ;-) --charleyb (all of them) [EMAIL PROTECTED]
-- http://cms-list.org/ more signal, less noise.
