Page MenuHomePhabricator

[SPIKE] Verify new source mode has not caused an increase in dirty diffs
Closed, DeclinedPublic


This task is about verifying DiscussionTools` new source mode (T276608) has not caused any regressions in the Reply Tool's Dirty Diff Rate.


Prior to the DiscussionTools` new source mode being made available to all users by default (read: as an opt-out setting) in T276609, it is important that we verify this new mode does not cause an increase in the dirty diff incident rate.

Dirty diff rate

Average dirty diff rate in 4 weeks preceding new source mode's deploymentAverage dirty diff rate in 4 weeks following new source mode's deployment (T276608)

//Data can be found in: DiscussionTools usage report


Event Timeline

I'll note that it's very unlikely that the new source mode will lead to more dirty diffs (changes elsewhere on the page when adding a new comment) – it adds the wikitext to the page in the exact same way as the old one.

However, it might lead to more mistakes in the wikitext that is added, for example because a user accidentally clicks a toolbar button, or because a user misunderstands the "Convert formatting to wikitext" dialog when pasting and ends up with a bunch of nowiki tags.

We don't have a way to automatically detect the latter.

ppelberg closed this task as Declined.EditedAug 2 2021, 5:57 PM

I am declining this task considering what Bartosz shared in T279396#6976348. Namely: the New Discussion Tool uses the same process for publishing topics drafted with the New Discussion Tool to talk pages as the existing experience does. And – to our knowledge – the existing &action=edit&section=new workflow does not cause content corruption.