Open asr opened 7 years ago
Using Equinox version 5.0alpha 2010-06-29 (commit 9151a09, compiled with GHC 6.10.4), the attached problem is proved instantly:
$ time equinox regression.fof Equinox, version 5.0alpha, 2010-06-29. +++ PROBLEM: regression.fof Reading 'regression.fof' ... OK +++ SOLVING: regression.fof #elt|m|#instances 12|P|.... 12|L|1.>. 99|P|.... 99|L|.L.1 190|P|L13. +++ RESULT: Theorem real 0m0.150s user 0m0.142s sys 0m0.008s
but using the upstream version the same problem is not proved in 10 minutes:
$ equinox --time 600 regression.fof Equinox, version 6.0.1alpha, 2011-12-07, pre-release. +++ PROBLEM: regression.fof Reading 'regression.fof' ... OK +++ SOLVING: regression.fof ... *** TIMEOUT (600 seconds) +++ RESULT: Timeout
Actually, I found the same kind of regression (proved in 5.0alpha 2010-06-29, not proved in upstream) in others TPTP problems, but the attached one is the simplest.
regression.fof.txt
In this branch, there is a version of Equinox 5.0alpha 2010-06-29 which should compile with recent versions of GHC.
CC'ing @jonaprieto.
Using Equinox version 5.0alpha 2010-06-29 (commit 9151a09, compiled with GHC 6.10.4), the attached problem is proved instantly:
but using the upstream version the same problem is not proved in 10 minutes:
Actually, I found the same kind of regression (proved in 5.0alpha 2010-06-29, not proved in upstream) in others TPTP problems, but the attached one is the simplest.
regression.fof.txt