[developers] Bug in mt/vpm.lisp

Dan Flickinger danf at csli.stanford.edu
Wed Mar 21 22:42:06 CET 2007


Thanks, Berthold and Bernd, for tracking down this bug - I had just run into
it this afternoon, and was trying to pin down the source of it.  

 Dan

> Today had me victim of a nasty bug in map-mrs():
> 
> With standard optimisation settings (:speed 3 :safety :2), generation
> with the ERG manage to make ACL dump core after a while. Increasing
> safety to 3 while decreasing speed to 1 (or 2), Bernd and I managed to
> track down the bug to the above function (error surfacing
> instantaneously): apparently, map-mrs assumes presence of CFROM and CTO
> features in MRS, only in this case they were not there (generated from
> Pet without characterisation). Uncommenting the two code lines referring
> to cfrom/cto solves the issue.  
> 
> 	       ;:cfrom (mrs:char-rel-cfrom ep)
> 	       ;:cto (mrs:char-rel-cto ep)
> 
> Probably map-mrs should check for the type of MRS coming in, or else
> presnece of these features. 
> 
> Cheers,
> 
> B
> 
> -- 
> Deutsches Forschungszentrum fuer Kuenstliche Intelligenz GmbH
> Trippstadter Strasse 122, D-67663 Kaiserslautern, Germany
> Geschaeftsfuehrung: Prof. Dr. Dr. h.c. mult. Wolfgang Wahlster (Vorsitzender)
, Dr. Walter Olthoff
> Vorsitzender des Aufsichtsrats: Prof. Dr. h.c. Hans A. Aukes
> Amtsgericht Kaiserslautern, HRB 2313
> 



More information about the developers mailing list