Open brennie opened 1 day ago
Test failure log: https://www.pastery.net/nbmzwm/
relevant section:
{'absolute': {'all': [],
E 'first': {}},
E 'difference': {'control': {'all': [],
E 'first': {}},
E 'variant': {'all': [{'lower': 0.0,
E 'point': 0.0,
E 'upper': 0.0},
E {'lower': 0.0,
E 'point': 0.0,
E 'upper': 0.0}],
E 'first': {'lower': 0.0,
E 'point': 0.0,
E 'upper': 0.0}}},
E 'relative_uplift': {'control': {'all': [],
E 'first': {}},
E 'variant': {'all': [],
E 'first': {}}},
E 'significance': {'control': {'overall': {},
E 'weekly': {}},
E - 'variant': {'overall': {'1': 'negative'},
E ? ---------------
E
E + 'variant': {'overall': {},
E 'weekly': {}}}},
The DataPoint.has_bounds implementation is:
which will report false for
self.lower = self.upper = 0
.Updating this check to
self.lower is not None and self.upper is not None
is probably more correct but breaks unit tests. (The log is too long to attach as a comment to this issue.)This happens because we end up in
compute_signficance
from here because has_bounds() is now true and the compute_signficance method doesnt expect lower = upper = 0┆Issue is synchronized with this Jira Task