Neither the mobile visual editor nor the mobile wikitext editor logs timing values for any of its events. The mobile editors do not log loaded events or, for the most part, abort events, so the others where the timing could be relatively easily added are:
| Subject | Repo | Branch | Lines +/- | |
|---|---|---|---|---|
| Include timing information for EditAttemptStep | mediawiki/extensions/MobileFrontend | master | +56 -2 |
| Status | Subtype | Assigned | Task | ||
|---|---|---|---|---|---|
| Resolved | nshahquinn-wmf | T202132 EPIC: Generate one-off metric snapshots for mobile editing documentation | |||
| Resolved | nshahquinn-wmf | T202147 Snapshot: which parts of the edit process cause people to abort edits? | |||
| Resolved | nshahquinn-wmf | T202437 Identify and fix data quality problems in the Edit event log | |||
| Resolved | nshahquinn-wmf | T205166 Mobile editors do not log timings for any Edit events |
Changing the parent tasks here since this doesn't block our immediate metrics work, since event timestamps are precise enough to calculate overall session duration.
I'm now changing the parent tasks back—as discussed in an Editing team meeting today, the lack of this data means we can't produce any metrics about the load performance of mobile editors, which will be crucial for our annual plan goal.
Change 476071 had a related patch set uploaded (by DLynch; owner: DLynch):
[mediawiki/extensions/MobileFrontend@master] Include timing information for EditAttemptStep
Change 476071 merged by jenkins-bot:
[mediawiki/extensions/MobileFrontend@master] Include timing information for EditAttemptStep
Almost the only QA possible for this one (besides it not actually breaking things) would be Neil saying whether the data being collected is correct, I think?
Everything looks good!
The results for the ready, saveIntent, saveAttempt, and abort events all look good. There are no results for loaded events because those are currently not logged (T206014). Timings are currently not logged for saveSuccess or saveFailure in any editor.
select
percentile_approx(event.ready_timing, 0.5) as median_ready_t,
count(event.ready_timing) / count(*) as nonnull_ready_t_prop
from event_sanitized.editattemptstep
where
event.platform = "phone" and
event.action = "ready" and
year = 2019 and month = 1
median_ready_t nonnull_ready_t_prop
1 714.7178125365697 0.9999889966604865select
percentile_approx(event.save_intent_timing, 0.5) as median_save_intent_t,
count(event.save_intent_timing) / count(*) as nonnull_save_intent_t_prop
from event_sanitized.editattemptstep
where
event.platform = "phone" and
event.action = "saveIntent" and
year = 2019 and month = 1
median_save_intent_t nonnull_save_intent_t_prop
1 51605.86 0.9999716312056738select
percentile_approx(event.save_attempt_timing, 0.5) as median_save_attempt_t,
count(event.save_attempt_timing) / count(*) as nonnull_save_attempt_t_prop
from event_sanitized.editattemptstep
where
event.platform = "phone" and
event.action = "saveAttempt" and
year = 2019 and month = 1
median_save_attempt_t nonnull_save_attempt_t_prop
1 5308.5 0.999962161344029select
percentile_approx(event.abort_timing, 0.5) as median_abort_t,
count(event.abort_timing) / count(*) as nonnull_abort_t_prop
from event_sanitized.editattemptstep
where
event.platform = "phone" and
event.action = "abort" and
year = 2019 and month = 1
median_abort_t nonnull_abort_t_prop
1 2647.5199871457917 0.9999838553438811select
percentile_approx(event.loaded_timing, 0.5) as median_loaded_t,
count(event.loaded_timing) / count(*) as nonnull_loaded_t_prop
from event_sanitized.editattemptstep
where
event.platform = "phone" and
event.action = "loaded" and
year = 2019 and month = 1
median_loaded_t nonnull_loaded_t_prop
1 NULL NULL