search

probcomp / BayesDB

A Bayesian database table for querying the probable implications of data as easily as SQL databases query the data itself. New implementation in http://github.com/probcomp/bayeslite
http://probcomp.csail.mit.edu/software/bayesdb/
Apache License 2.0
891 stars 52 forks source link

skippable missing schema alerts and initialize time and fix HIST on nans #52

Closed fiendish closed 10 years ago

fiendish commented 10 years ago

change 1: Move missing schema alert from engine to client. Having the alert in engine makes it unskippable, which disrupts the utility of setting yes=True like on every other command that asks for confirmation. Moving it to client makes it so that yes=True bypasses the alert.

change 2: model['time'] is not always initialized in the persistence layer before adding to it.

probcomp-gh-bot commented 10 years ago

Can one of the admins verify this patch?

fiendish commented 10 years ago

change 3: numpy.histogram can't deal with NaNs, so block those out when using the HIST keyword. Otherwise if row 39 of column A is a nan, then "hist select A from table limit 39" will work, but limit 40 will break and return garbage.

BaxterEaves commented 10 years ago

ok to test