Re: Oddness...
Thomas Davie <[email protected]> Mon, 18 Jul 2005 16:18:49 +0000 (UTC)
| Newsgroups | gmane.comp.lang.haskell.hat |
|---|---|
| Message-ID | <403F2B7B-77BA-46DE-8901-86E49A63D26B__28997.5334344946$1121703529$gmane$org@kent.ac.uk> |
On 18 Jul 2005, at 16:58, Malcolm Wallace wrote: > Thomas Davie <[email protected]> writes: > > >> I've been trying to get hat-trans to eat QuickCheck, without much >> success. Is there something I'm missing? I'm getting this error: >> >> hat-trans QuickCheck.hs >> hat-trans: Variable or constructor not in scope: chr >> > > The QuickCheck.hs module available from John Hughes' webpage was > written for an older version of Hugs which wasn't compliant with > Haskell'98. The 'chr' function is exported from the standard Char > library. > > However, I have to wonder whether it makes any sense to trace > QuickCheck itself? In the AFP summer school 2002, the students used > QC to find a failure, then fed only that failing case into Hat to > discover its cause. > > For delta-debugging, I expect you would want both a failing case and > a passing case, but I don't suppose you really want to bother with > tracing all the QC machinery that found them for you. True, this was really an attempt at a first hack that involved running quickcheck (and thus getting multiple successes and failures that are easy to identify)... Look for evaluations of function starting prop_ and use them to do some delta debugging... Before I start accepting multiple traces, etc. Bob