[Cdk-devel] jjtree my comment

[sushil r. <ron...@gm...>]

I think the present smartparer we have is  doing very well in all aspect.
On performance scale: for the biggest molecule like Pioglitazone it took 1
second.

Coding into jjtree will not add any benifit over present parser.
Except the loss of energy of the few administrator we have at present.



> What I added are parsing logics. So far the JavaCC code in SVN doesn't
> handle logical operators, recursive smarts, etc. Maybe Sushil has
> already implemented those locally as well, then we can exchange ideas of
> how we did these and merge our code.
>
>
you are right i am ready with the operator code.(but at present we need some
attention on UIT and SQT,(query API))
and present parser is also do the recursion except for operators at present.

So i think coding the smartparser again in jjtree is certainly a loss
of energy at both side.


> What makes the JJTree approach good is, people who are good at parsing
> can focus on parsing,
>

that is not the right attitude i think,making smart parser is not a big deal
it has atoms and bonds which are already a node
and we are making a tree of it in the atom container.


But on contrary CDK certainly need a experience programmer like you.
There are  some other aspect where cdk certainly need a help.
Any contribution in that regard will certainly add a value to cdk.


-- 
********************************
sushil ronghe
Center of Pharmacoinformatics
India
*********************************