- Mar 23, 2016
-
-
ravi authored
-
- Mar 22, 2016
-
-
ravi authored
(b) Fixing bugs in the benchmarks (c) Makinging transformation phases of Orb ignore parts of the library
-
Etienne Kneuss authored
* Expression grammars are now alwayse 'Labels -> Expr', simple grammars that are 'Type -> Expr' can simply extend SimpleExpressionGrammars * Labels now have Aspects. A label aspect describes a particular feature of the label (e.g. its size), as well as how this aspect applies to sub-productions. Aspects describe how sub-productions are generated/filtered/modified. See comments in Aspect for more information
-
- Mar 17, 2016
-
-
Regis Blanc authored
-
Manos Koukoutos authored
-
Manos Koukoutos authored
-
Manos Koukoutos authored
terminatingCalls now returns calls with integers too. IntInduction and all integer comparison rules are phased out. Instead there is a single InequalitySplit rules which splits into (up to) 3 branches, taking pc into account. No more EqualitySplit for any other type, except generic types. Isolate unused rules. Clean up Rules. Improve QualifiedExamplesBank API
-
Manos Koukoutos authored
-
Manos Koukoutos authored
-
Manos Koukoutos authored
-
Manos Koukoutos authored
More reasonable return type for solveSATWithCores CEGIS discards tests that crash the evaluator Try harder to find verifiable solution in validatePrograms Increase default CEGIS size
-
Etienne Kneuss authored
-
Manos Koukoutos authored
Additional fixes: - move OneOf to grammars - Propagate Terminating in IntroduceRecCall
-
Manos Koukoutos authored
Remove SafeRecursiveCalls from Grammars.default IntroduceRecCalls keeps only examples that lead to no 'choose' IntroduceRecCalls is normalizing Make 1 benchmark tougher
-
- Mar 16, 2016
-
-
Romain Edelmann authored
See Stack Overflow question: http://stackoverflow.com/questions/32022588/why-the-wrongcommutative-timeout-in-the-propositionallogic-example-in-leon-onlin
-
- Mar 15, 2016
-
-
Regis Blanc authored
-
Regis Blanc authored
-
- Mar 08, 2016
-
-
ravi authored
-
Mikaël Mayer authored
Simplified the double list benchmark which works now.
-
- Mar 07, 2016
-
-
ravi authored
-
Mikaël Mayer authored
-
- Mar 06, 2016
-
-
Ravi authored
Minor changes MInor changes
-
- Mar 05, 2016
-
-
ravi authored
(b) Preserving if-then-elze in the VCs (c) Optimizing elimnation and UFADT reduction
-
- Feb 28, 2016
-
-
Nicolas Voirol authored
-
Nicolas Voirol authored
-
- Feb 15, 2016
-
-
Ravi authored
-
- Feb 10, 2016
-
-
Mikaël Mayer authored
-
- Feb 09, 2016
-
-
Regis Blanc authored
-
- Feb 08, 2016
-
-
Manos Koukoutos authored
-
Manos Koukoutos authored
-
- Feb 05, 2016
-
-
ravi authored
-
Manos Koukoutos authored
ValueGrammar only generates one generic value. GrammarDataGen will expand to all non-redundant value combinations. Use GrammarDataGen instead of bonsai everywhere in Leon (except TEGISLike)
-
- Feb 04, 2016
-
-
ravi authored
-
Mikaël Mayer authored
-
- Feb 03, 2016
-
-
ravi authored
-
- Feb 02, 2016
-
-
ravi authored
-
Mikaël Mayer authored
Question builder will now try to instantiate always different terminal values for better comparison. Corrected benchmark out of order to not depend on each other;
-
Marco Antognini authored
And remove some extra whitespace
-
ravi authored
-
- Feb 01, 2016
-
-
ravi authored
-