You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
The all-fire context is computed for each determinized operator (unlike the conditional mask, which is computed for the full non-det operator), and it's only used for the regression of generalized deadends. It was a minor hack (unpublished) that let us gain a few more FSAPs in certain corner cases, and worked when the conditional effects were not nested within the oneof clauses. It's current implementation is incorrect when the condeffs are inside the oneof, as above -- result is an FSAP potentially being created that is too "strong" (rules out things it shouldn't).
The text was updated successfully, but these errors were encountered:
This code should be disabled for now, and then rethought: https://github.com/QuMuLab/pr2/blob/main/src/deadend.cc#L120
It's rarely used (requires a domain with conditional effects), and actually is buggy in the following situation:
The all-fire context is computed for each determinized operator (unlike the conditional mask, which is computed for the full non-det operator), and it's only used for the regression of generalized deadends. It was a minor hack (unpublished) that let us gain a few more FSAPs in certain corner cases, and worked when the conditional effects were not nested within the
oneof
clauses. It's current implementation is incorrect when the condeffs are inside theoneof
, as above -- result is an FSAP potentially being created that is too "strong" (rules out things it shouldn't).The text was updated successfully, but these errors were encountered: