-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Michael Parker writes:
> John Myers wrote:
> > The rule promotion criteria don't appear to be doing the right thing
> > with the DATE_IN_* tests.
> > 
> > These tests are essentially a group.  Once one of them has been paid
> > for, the rest have negligible incremental cost.
> > 
> > DATE_IN_FUTURE_48_96 is disabled due to lack of hits, but the tests on
> > either side of it are enabled.  Most of the DATE_IN_PAST_ tests are
> > disabled due to S/O ratios below the cutoff.
> > 
> > I believe all of these tests should be active.  I'm not sure how to
> > generalize this observation to other sets of eval tests.
> 
> Maybe we should put rules into groups or buckets and the rule promotion
> code would add all rules in that group/bucket when any of them are selected.

Hold on, though.

If the S/O ratios are too poor to qualify, why bother keeping them, even
if they *do* have negligible cost?   They still have a cost, even if it is
small; the overhead of calling an eval rule, and tracking
scores/descriptions/etc.

- --j.
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.1 (GNU/Linux)
Comment: Exmh CVS

iD8DBQFDutxoMJF5cimLx9ARAulDAJ9qUWf7WD2QSAuSoemyYYBJIyPpfgCgp4Ao
xa1Hl5hvCX41sZZSkS2axCk=
=AaFd
-----END PGP SIGNATURE-----

Reply via email to