cclib-devel

Re: [cclib-devel] how I use cclib

[Noel O'B. <bao...@gm...>]

That's nice. Maybe we can put some examples of extending cclib up on
the wiki. BTW, why do you extract natoms yourself?

On 06/06/07, Karol Langner <kar...@kn...> wrote:
> Hi,
>
>  I thought I might share with you how I use cclib in my everyday work now. In
> our lab we have a custom modified version of GAMESS, and I analyze hundreds
> of output files each week. Before, I had to write parsers from scratch to
> gather data. With cclib, it's as easy as writing a class that inherits from
> the GAMESS parser in cclib. See the attached file for an example!
>
> Cheers,
> Karol
>
> --
> written by Karol Langner
> Wed Jun  6 09:51:43 CEST 2007
>
>

Re: [cclib-devel] how I use cclib

[Karol L. <kar...@kn...>]

On Wednesday 06 June 2007 10:37, Noel O'Boyle wrote:
> That's nice. Maybe we can put some examples of extending cclib up on
> the wiki.

That's a good idea. This one isn't readily useable by everyone, though, since 
it's for our internal lab code.

> BTW, why do you extract natoms yourself? 

I need to do this, because each EDS output file actually contains three 
calculations - for each monomer and the dimer (like in Morokuma 
decomposition). Every time the nubmer of atoms is different. There are other 
things that should also be modified, but I don't need them for now so I 
haven't done that.

- Karol

-- 
written by Karol Langner
Wed Jun  6 21:18:40 CEST 2007

Re: [cclib-devel] how I use cclib

[Karol L. <kar...@kn...>]

There is another thing I wanted to share/discuss. I have a little thing I use 
to store my output data in XML files, which is something that was brought up 
before. It's not compatible with cclib in its present form, since I have been 
using it with scripts I wrote before I knew about cclib. I would like to base 
it on cclib objects, though. When I do this, maybe it could be a starting 
point for adding a similiar functionality to cclib.

- Karol

-- 
written by Karol Langner
Wed Jun  6 21:25:58 CEST 2007

Re: [cclib-devel] how I use cclib

[Noel O'B. <bao...@gm...>]

On 06/06/07, Karol Langner <kar...@kn...> wrote:
> There is another thing I wanted to share/discuss. I have a little thing I use
> to store my output data in XML files, which is something that was brought up
> before. It's not compatible with cclib in its present form, since I have been
> using it with scripts I wrote before I knew about cclib. I would like to base
> it on cclib objects, though. When I do this, maybe it could be a starting
> point for adding a similiar functionality to cclib.

I see that resistance is futile. :-) Here are some ideas:
(1) pprint the attributes to a file. This can be evaluated in Python.
(2) JSON it to a file (http://www.json.org/). This can be read by
several languages (apart from Javascript), and is great for
webservices (e.g. we can put a webservice on SF). In fact, according
to http://effbot.org/zone/delicious-json.htm it can simply be
evaluated in Python.
(3) Don't use XML. Use something like Gaussian formatted checkpoint,
but with comments in the file that describe what the data in the file
is. E.g. the file format would be self-describing with units,
descriptions of the data, description of the data format.

> - Karol
>
> --
> written by Karol Langner
> Wed Jun  6 21:25:58 CEST 2007
>

Re: [cclib-devel] how I use cclib

[Karol L. <kar...@kn...>]

On Tuesday 12 June 2007 18:45, Noel O'Boyle wrote:
> On 06/06/07, Karol Langner <kar...@kn...> wrote:
> > There is another thing I wanted to share/discuss. I have a little thing I
> > use to store my output data in XML files, which is something that was
> > brought up before. It's not compatible with cclib in its present form,
> > since I have been using it with scripts I wrote before I knew about
> > cclib. I would like to base it on cclib objects, though. When I do this,
> > maybe it could be a starting point for adding a similiar functionality to
> > cclib.
>
> I see that resistance is futile. :-) Here are some ideas:
It's just an idea :) cclib has already made my processing output more 
efficient, so I am willing to contribute more to it.

> (1) pprint the attributes to a file. This can be evaluated in Python.
> (2) JSON it to a file (http://www.json.org/). This can be read by
> several languages (apart from Javascript), and is great for
> webservices (e.g. we can put a webservice on SF). In fact, according
> to http://effbot.org/zone/delicious-json.htm it can simply be
> evaluated in Python.
> (3) Don't use XML. Use something like Gaussian formatted checkpoint,
> but with comments in the file that describe what the data in the file
> is. E.g. the file format would be self-describing with units,
> descriptions of the data, description of the data format.
All good points. Of course, i will probably be a while before I have the time 
actually get into this.

-- 
written by Karol Langner
Tue Jun 12 22:11:30 CEST 2007