1106617 - PLACES_AUTOCOMPLETE_6_FIRST_RESULTS_TIME_MS histogram regressed the 25/11

Status: RESOLVED FIXED

(Toolkit :: Places, defect, P3)

Description

[Roberto Agostino Vitillo (:rvitillo)]

Please see http://alerts.telemetry.mozilla.org/index.html#/detectors/1/metrics/957/alerts/?from=2014-11-25&to=2014-11-25

Comment 1

[Marco Bonardo [:mak]]

I thought this was when we disabled unifiedcomplete but that happened on 26 (see bug 1105456)... So this could have 2 reasons:
1. could be the fact now for the FIRST search in a session we must fetch keywords, though that patch has been backed out for now
2. could be pre-selecting the first entry somehow caused more reflows

Unfortunately it's very hard to say now, cause we had to disable unified on 26 and now the histogram is measuring OLD autocomplete code...

Comment 2

[Roberto Agostino Vitillo (:rvitillo)]

Also PLACES_AUTOCOMPLETE_1ST_RESULT_TIME_MS regressed: http://alerts.telemetry.mozilla.org/index.html#/detectors/1/metrics/700/alerts/?from=2014-11-27&to=2014-11-27

Comment 3

[:shell escalante]

check again after enable unifiedautocomplete in nightly

Comment 4

[Marco Bonardo [:mak]]

According to the telemetry dashboard both PLACES_AUTOCOMPLETE_1ST_RESULT_TIME_MS and PLACES_AUTOCOMPLETE_6_FIRST_RESULTS_TIME_MS improved with current version of unified complete (builds >= 30th of May)
TBH, we should now measure the second entry, but still the scope of the probe was responsiveness, and we increased the perceived responsiveness.