Thread: Re: [cclib-devel] contribute to cclib

SourceForge Headquarters 1320 Columbia Street Suite 310 San Diego, CA 92101 +1 (858) 422-6466

Hello Mehdi,
> 
> I would like to contribute to cclib by writing a parser for
> Molpro (in fact I already started).
> 
Fantastic!

> How can I do that ?
This needs a longer answer. :-)

(1) Have you worked on an open source project before? If not, it might
be a good idea to skim read http://producingoss.com/ .
(2) Have you used SVN before? This is how we store our code. If not, you
might like to read chapters 1-3 of http://svnbook.red-bean.com/ .
(3) To start, check out the latest copy of our code as described under:
http://cclib.sourceforge.net/wiki/index.php/Development#Checking_cclib_out_of_subversion

Adam and myself will need to discuss the addition of a new developer. In
the meanwhile please still continue working on the parser, and we will
be able to merge the changes. Once you are added as a developer, we'll
look over your shoulder for a while as you commit changes to give you
some advice.

In relation to adding a new parser:
(4) We don't have access to Molpro, so you will have to run all of the
test jobs. Look at the 'data' folder on subversion and you will see that
for every comp chem package we have test jobs on divinyl benzene at the
B3LYP/STO-3G level of theory:
(a) a geometry optimization (with SCF convergence information and
geometry optimization convergence information)
(b) a single point energy calculation (with ao overlaps and mo coeffs)
(c) a single point unrestricted energy calculation (with ao overlaps and
mo coeffs)
(d) a vibration frequency calculation with IR information
(e) a vibration frequency calculation with Raman information (if
possible)
(f) a TD-DFT calculation (if possible)

Any additional calculations used to define the parser should not be
added to SVN - there is a different procedure, which I will describe
when you want to do this.

(5) The only way we stay sane is tests. No parser is released until it
passes all of the tests (run testall.py). You will need to add molpro to
the test scripts of course. This implies that every parser needs to
parse the same things, and use the same units (see the wiki for
information).

> Also have you considered parsing also first/second order
> derivative infomation ? 
I would be interested in hearing more about this - for example, how can
this information be used? Our goal with cclib is not to extract all
possible information, but that information which is useful for
algorithms (which cclib also includes) or visualization (by some other
software).

Regards,
 Noel

P.S. Please cc everything to the cclib developers list (usually by
clicking Reply All)