Bugzilla – Full Text Bug Listing

Summary: | inconsistency: Expression of type (constraint) cannot be used as a guard | ||
---|---|---|---|

Product: | UPPAAL | Reporter: | Timothy Bourke <tbourke> |

Component: | GUI | Assignee: | Alexandre David <adavid> |

Status: | RESOLVED FIXED | ||

Severity: | trivial | CC: | marius |

Priority: | P2 | ||

Version: | 4.0.6 | ||

Hardware: | PC | ||

OS: | Linux | ||

Architecture: |

Description
Timothy Bourke
2007-10-03 02:21:40 CEST
The code is quite scattered but formally the typechecker seems to implement the following rules: invariant ::= integral | (clock "<" clock) | (clock ">" clock) | (clock "<" integer) | (integer ">" clock) | (difference "<" integer) | (integer "<" difference) | (difference ">" integer) | (integer ">" difference) | (invariant "&&" invariant) | (integral "||" invariant) ; guard ::= invariant | (integer "<" clock) | (clock ">" integer) | (guard "&&" guard) | (integral "||" guard) ; constraint ::= guard | (clock "!=" clock) | (clock "!=" integer) | (integer "!=" clock) | (difference "!=" clock) | (difference "!=" clock) | "!" constraint | "forall" constraint | "exists" constraint | (constraint "&&" constraint) | (constraint "||" constraint) ; Typechecker will try to interpret the expression in simplest possible way (i.e. expression will be invariant if it matches invariant rule, and expression will get promoted to guard only if it cannot match invariant anymore and so on). Eventually some expressions might end up as constraints in general, which means that they are unsupported, they are easily recognized (type CONSTRAINT) and raise "cannot be used as" errors. As you can see from the rules, the reported problem persists also for invariants. Having said that, the following additional rules enable typechecker to accept the above expressions: invariant ::= | (invariant "||" integral) guard ::= | (guard "||" integral) This fix has been applied in rev. 3546. Behold the rest of the code :-) Thanks Marius. I also appreciate your inclusion of the rules. |