Re: [Cdk-devel] question about HybridizationStateATMatcher

[Rajarshi G. <rx...@ps...>]

On Sat, 2005-07-16 at 00:07 +0200, Christoph Steinbeck wrote:

> I would actually dare to state that the atomtype problem is the most severe in 
> CDK. And this is due to the fact that fixing it for working on 99% of the cases, 
> would involve a lot of work, but getting it to run for my current problem is 
> easy. That is kind of the fundamental problem of Open Source in small communities.
> 
> But anyway, I think your message was received, and it was more about 
> documentation than about non-working code. This is very much appreciated.

Thats correct - as I said, I have no problem with trying to fix code
that does'nt work. 

I realize that documentation is boring (and many times I've really had
to force myself to write up Javadocs!), however the problem is not so
much for regular developers on the list, who have a general idea of
whats going on. 

For a developer who needs cheminformatics functionality and turns to the
CDK, undocumented features/limitations/todo's etc all detract from the
quality of the code.

I'm in line with Joerg's view (mentioned before on this and other lists)
that we need to consider cheminformatics developement as a software
engineering situation. And hence, we need some rigor. I think Egons
proposal of a QA team and the limitations module are the beginnings of
this type of approach.



-------------------------------------------------------------------
Rajarshi Guha <rx...@ps...> <http://jijo.cjb.net>
GPG Fingerprint: 0CCA 8EE2 2EEB 25E2 AB04 06F7 1BB9 E634 9B87 56EE
-------------------------------------------------------------------
All science is either physics or stamp collecting.
-- Ernest Rutherford