New obthermo tool with corresponding data

[dspoel]

This provides a new tool that extracts thermochemistry (Delta H formation etc.) from a Gaussian log file. The patch also provides the necessary reference data from a large set of calculations and experiments.

[ghutchis]

Dunno, there seems to be an issue with appveyor. Can you try "bumping" this with a new update?

[ghutchis]

It looks OK to me, although Windows builds use a very different compiler/linker.

[dspoel]

Could it be that the file data_utilities.cpp is not included in the build on Windows? It is in the CMakeLists.txt file.

[baoilleach]

The problem is that the functions in data_utilities are not exposed from the openbabel DLL. On Linux, by default, everything is exposed, while on Windows you have to specify which ones to expose.

More to the point, we tend to avoid extending the OB API unless necessary. I'd suggest you simply move the data_utilities.cpp out of the openbabel core and into the thermo target. Obviously there's no linking problem then.

However, if you can make a case for keeping them where they are (e.g. are these functions of general use to OB users?), I'm happy to do the necessary to expose the functions.

[dspoel]

Thanks for your comments. One of the new routines energyToKcal(std::string) is used by gaussformat.cpp to convert units to kcal, this in order to remove the shattered physical constants in the code. I can remove the call and put the constants back if you think that is better.

[dspoel]

In fact I would like to use the function extract_thermochemistry from my GROMACS programs, this is why I implemented all this, so I would be interested in an API change. If there are better ways I am open for suggestions though.

[baoilleach]

Sorry it's taking me a while to respond.

Could energyToKcal be implemented instead of a set of consts in the header file (e.g. "const double JOULE_TO_CAL = whatever")? This wouldn't need any API changes though we could expose it if useful. We have something similar for DEG_TO_RAD I think. The nice thing about a const is that the IDE will fill in the name of the variable, though I guess a function taking an enum would do the same.

I'll do the necessary work for extract_thermochemistry tomorrow. Hopefully I can send you a pull request directly, and then you get to hassle me about my patch! :-)

[baoilleach]

Can you email me a test file and associated command-line. I've tried a few files, but nothing I have seems to trigger the code.

[dspoel]

On 28/07/15 17:20, baoilleach wrote:

Can you email me a test file and associated command-line. I've tried a
few files, but nothing I have seems to trigger the code.

—
Reply to this email directly or view it on GitHub
#200 (comment).

obtherm butane-g4.log.gz
should do the trick.

David van der Spoel, Ph.D., Professor of Biology
Dept. of Cell & Molec. Biol., Uppsala University.
Box 596, 75124 Uppsala, Sweden. Phone: +46184714205.
spoel@xray.bmc.uu.se http://folding.bmc.uu.se

[dspoel]

http://folding.bmc.uu.se/images/butane-g4.log.gz

[baoilleach]

Thanks - you should have received a pull request for my code. If you accept this, I presume it will magically appear here.

[dspoel]

Cool, thanks! We're about to submit the paper on this.