Sorry for the confusion, it turns out the user reporting the problem was copy/pasting incomplete InChIs ... I wish InChI package had less cryptic error reporting.
Still ChemSpider searching seems to be only one to cope with incomplete InChI - is it possible with (any) InChI library to achieve something similar - e.g. extract only a skeleton from available InChI layers ?
Attemping to parse this InChI=1S/C6H8O6/c7-1-2(8)5-3(9)4(10)6(11)12-5/h2 via org.openscience.cdk.inchi.InChIToStructure (latest stdinchi ) fails with an error:
Aux. info suppressed
Input format: InChI (plain identifier)
Output format: Plain text
No timeout; Up to 1024 atoms per structure
Structure: 1 Syntax error (-2) in MOBILE_H (2)
1) OpenBabel also have troubles parsing this string.
2) http://rdf.openmolecules.net/?InChI=1S/C6H8O6/c7-1-2%288%295-3%289%294%2810%296%2811%2912-5/h2 links to water (which is obviously wrong)
3) ChemSpider seems to be the only site (among I've tested ) which succeeds in finding the compound
Am I missing something? Is this a known issue ?