Geoffrey Hutchison wrote: > Currently, the InChI test is failing on UNIX platforms for reasons I can't > understand. Here's a run-down: > http://my.cdash.org/testDetails.php?test=1930810&build=58657 > > So the main problem is that the molecule (ferrocene) shows an error: > > Not ok 1 # Mismatch in molecule #1. generated / correct > # InChI=1S/2C5H5.2C5H.2Fe/c4*1-2-4-5-3-1;;/h2*1-5H;2*1H;;/q;;;;;+2 > # InChI=1S/4C5H5.2Fe/c4*1-2-4-5-3-1;;/h4*1-5H;;/q;;2*-1;;+2 > > OK, so the second InChI is the correct one. But if you run "make install" and > then run babel to generate the inchi from the SDF yourself, you get that > *exact* InChI. So there's no problem with our code. > > I'm left trying to figure out the problem with the test environment. > * Chris: what data files could mess up the test? > * Marcus: how would CTest differ from a compiled babel?
This may be clutching at straws, but the file samples.sdf has Windows line endings. This should not matter with either inchiwrite.cpp or babel, but possibly something has gone wrong. I've committed a version with Unix line endings, so you can try it. It is just about possible that this would affect the parsing of the sdf 'alias' data, which is where you suggest the problem is. The InChI sdf samples use alias constructs like A 1 CH A 4 CH- which is not good practice in my view; there are better ways of saying that the C has an H and a maybe a negative charge. The code which interprets this construct, AliasData::FormulaParse, was cobbled together so that these InChI sample files could be read, and I would have little confidence in its robustness. I was not inclined to write a (much larger) complete parser for inline formulas, because I am not aware of a specification for it, and anyway construct is obsolete and IMO unnecessary. (Defined aliases like COOH are much more sensible and are supported.) Chris ------------------------------------------------------------------------------ Download Intel® Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev _______________________________________________ OpenBabel-Devel mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/openbabel-devel
