cocoon-users mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Kevin Sonney <ke...@webslingerZ.com>
Subject RE: Storing XML in databases w/o lots of string parsing....
Date Fri, 21 Apr 2000 17:54:12 GMT
On Fri, 21 Apr 2000, Williams, Murray Todd wrote:
> But if you are doing searches using the SQL "like" keyword, you are still
> storing the data in a traditional ASCII representation of XML, right?  So
> you still use a standard SAX or DOM parser to convert the ASCII stream into
> a DOM document or series of SAX events, right?

Yeah, that was the drawback. We'd have the full XML doc stored in the
CLOB, search using Like, and after narrowing down the 1000+ (from 10000+)
documents we have to go back and parse them to find out if the keyword
we're after is in the right element. 

Now one option that just occured to me, is if you *KNOW* the element
you're seaching for, and the keyword, you could use "where XMLCLOB like
'%<element>%keyword%</element>%' ", narrowing your 10000+ documents down
to the 5 or so with <element>...keyword...</element> somewhere in the
text. This wouldn't work with attributes to an element, though. But it
would simplify searching, say, huge quantities of documents with important
content - like XHTML bodies and such.

> Well, you can still help by pointing out where my misconceptions of XML,
> parsers, serialization and caches might fall.  If you have any insight,
> please let me know.

Honestly, I don't really understand 'em all that well myself. I'm more of
a data storage/JDBC kind of guy. Mostly. 

Although I *DID* just get something really cool along the lines of :

LDAPProcessor -> XSLT -> XSP -> SQL Taglib -> XSLT -> HTML

working. Now if only I can make it work like my customewr wants it to...

-- 
- Kevin Sonney
  kevin@webslingerZ.com


Mime
View raw message