mozilla / active-data-recipes

A repository of various activedata queries and recipes
Mozilla Public License 2.0
9 stars 24 forks source link

backout_rate and try_efficiency rate scripts return incorrect values due to ActiveData API only returning 10000 rows #136

Open Archaeopteryx opened 5 years ago

Archaeopteryx commented 5 years ago

The ActiveData API returns maximum 10000 rows (confirmed by ekyle) but some queries request more, e.g. 100000. That causes incorrect calculations.

E.g. the backout rate gets currently calculated by getting all push ids, generating a set from the result for deduplication (ekyle likely can explain why there are duplicates) and getting its length. So for queries over a longer time frame (e.g. since start of the year), the total push count gets capped, bloating the backout rate.