On Fri, 22 Mar 2002, Elias Karakoulakis wrote: > I have opened a bug case in bugzilla for the ESQL logicsheet: it doesn't > behave correctly for some kinds of grouped resultsets (<esql:group>). I > have tried to debug the logicsheet but it's far too complex to understand > (yet alone debug...) Suffice to say that most logicsheets in the same dir > are about 10-20kb in size, and esql.xsl is about 55kb (plus the size of > the accompanying Esql*.java classes)
the esql logicsheet implementation can be considered to be an example of the limits of xslt-generated source code. i don't think it's too complex to understand (although i don't know how much it's mutated since i left it), but debugging is another matter. that being said, i don't think it's possible to refactor the esql logicsheet to act more like the other logicsheets (that is to say, simply providing an xml api to a static library or object get/set methods) without losing the ability to execute arbitary code (via xsp tags, other logicsheet tags) based on the results of the queries. and that, as far as i can tell, is the only good reason to choose esql over the sql transformer. > A rather daunting question: would it be wise for use to be thinking about > reimplementing the ESQL logicsheet, just to scale down its complexity? I > have rewritten the xsd schema, simplifying the <esql:get-*> > elements.. (really, why did there had to be a separate element for each > datatype?). Another big problem is that of poor documentation: Except from > the userdoc "howto", there is very little to go from there.. I think that > starting from a properly documented schema would be of major importance. there was a seperate element for each datatype because i was trying to mirror the jdbc api. i don't know that that was a good reason, but that's why. anyway, i don't have specific comments on your proposal, i just happened to see your note and thought i'd dash off a response before i forgot about it. good luck in your efforts. - donald --------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, email: [EMAIL PROTECTED]
