[pet] PET code cleaning
Peter Adolphs
Peter.Adolphs at dfki.de
Thu Jul 8 11:42:26 CEST 2010
Hi!
On 06/07/10 13:24, Bernd Kiefer wrote:
> To all: to do the "more serious refactoring", could we manage to
> put up a system/procedure that allows for thorough regression testing
> in short time? That does not need to be THE testing environment we
> envisage in the future, but it should make relatively sure that there's
> nothing broken.
Do you mean unit tests or regression tests on the parser's output?
As to the latter, I use the [incr tsdb()] from the logon distribution for that
purpose. I usually parse Cathedral&Bazaar with the erg+tnt profile (that
includes chart mapping). So this is what I'd recommend to do:
0) Create a modified CPU definition for your PET version, say "erg+tnt at mycheap".
1) Create a reference profile, e.g. using the PET version coming with LOGON:
cd $LOGONROOT
./parse --binary --erg --client erg+tnt --best 500 --suffix .logon cb
2) Create the profile for your new PET version:
cd $LOGONROOT
./parse --binary --erg --client erg+tnt at mycheap --best 500 --suffix .mine cb
3) Compare the two profiles in [incr tsdb()]:
- middle-click the reference profile
- right-click the new profile
- Compare -> Competence / Performance
Cheers,
Peter
--
Peter Adolphs
DFKI GmbH, Projektbuero Berlin, Alt-Moabit 91c, D-10559 Berlin
Telefon: 030 / 3949-1826
_______________________________________________________________
Deutsches Forschungszentrum fuer Kuenstliche Intelligenz GmbH
Firmensitz: Trippstadter Strasse 122, D-67663 Kaiserslautern
Geschaeftsfuehrung:
Prof. Dr. Dr. h.c. mult. Wolfgang Wahlster (Vorsitzender)
Dr. Walter Olthoff
Vorsitzender des Aufsichtsrats:
Prof. Dr. h.c. Hans A. Aukes
Amtsgericht Kaiserslautern, HRB 2313
More information about the pet
mailing list