https://bugzilla.wikimedia.org/show_bug.cgi?id=37463
--- Comment #15 from Quim Gil <q...@wikimedia.org> --- Status update (Nemo, I will leave your comment to Alvaro since I don't know better) We have some graphs here: http://korma.wmflabs.org/browser/gerrit_review_queue.html "How long is the Gerrit review queue over time?" and "Backlog of pending reviews (when reviews were submitted)" could be put together, both starting when Gerrit was set up (march 2012?). Maybe they could even be combined in a single graph? (optional) Little known bug: all graphs should have Y axis starting with 0. "How long does it take to review code contributions?" and "Are Wikimedia's staff and non-staff contributions processed equally?" also belong to each other. At the end the second question equals "How long does it take to review code contributions... from WMF, independents, WMDE, Wikia...?" We are still discussing what to count exactly here. We started the discussion in our weekly meeting and we will continue here. The last assumption (which is not reflected in the graphs now) is: For each measurement timespot, median of the ages of all reviews still open at that timespot, understanding as "age" the time period since the review was open to the measurement timespot. Note that a review can have multiple patches. I'm proposing to count the age of the last patch instead of the age of creation of the review. What we are most interested in seeing is lack of activity, languishing reviews. Counting the age since creation is also useful, as secondary value (a second graph?) "Who is doing better and worse?" should be sorted by median age of open reviews, starting by those repos with oldest patches awaiting review. ------------------------ In our discussion there were more ideas about data we could count. I'm documenting them here, although this might be relegated to a second iteration (since we have other KPIs that we want to complete in the following weeks): * median time between uploading the changeset and receiving a first review * average of iterations vs time since creation (some changesets may take long to review but they were nailed in a couple of revisions as soon as someone answered, while others required many iterations) * normalization of data for big projects, calculating number of revisions within a period compared with the previous period. -- You are receiving this mail because: You are on the CC list for the bug. _______________________________________________ Wikibugs-l mailing list Wikibugs-l@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/wikibugs-l