[Cdk-devel] an mine too (was: my view)

[Egon W. <e.w...@sc...>]

On Monday 23 April 2007, sushil ronghe wrote:
> > From your mails Sushil I get the feeling that you fight against the other
> > solution, but there is no reason for this
>
> well if you feel so i must have written some thing of that kind.
> Your view is very correct boss.

This are my observations:

1. Sushil puts forward not just an architecture, but a full engine
2. he shows a lot of commitment
3. the JJTree option has potential (*), but does not allow actual searching 
yet

*) for example for optimization

Combine this with basic opensource economics:

a. development is done by people who actually write code
b. code evolves

Now, regarding a), Sushil is currently to only one who is extending SMARTS to 
the full language, and as primary developer on that, his prerogative to 
choose the method. His patch actually does something, so will go into the CDK 
as he likes it.

Regarding b), if, at some point, the JJTree approach allows doing actual 
searches too, it can be merged. That could have happened from the start. But 
it did not, possibly because people start hacking first, and only then 
discuss things on the mailing list.

Anyway, a developer has the obligation for respond to bug reports, nightly 
reports, etc, but has the freedom to provide a patch of any architecture he 
likes. Other developers may disagree with that architecture, but then, in 
turn, have the obligation to write patches. That is, you can not expect 
others to implement ideas you have (unless you pay them).

Applied to this situation, it would actually help if the JJTree code was 
available and fully integrate with what Sushil has been doing. That is, to 
speak with Linus: "Show me the code!". That's really what it comes down too.

That's my view on this.

Egon

PS. regarding (SMARTS) query optimization, the AST approach is promosing here. 
I suggested earlier that the to approaches can complement each other here: 
the AST can optimize the SMARTS which can then be serialized to a string and 
then send to Sushils parser.

-- 
e.w...@sc...
Blog: http://chem-bla-ics.blogspot.com/
GPG: 1024D/D6336BA6