Wed, Mar 3
FWIW I don't think it's necessary to revert the patch that I merged today. @Jdlrobson left some helpful comments on the patch that should get us to the solution 🙂
Being bold. I re-ran the tests and it passes. I've also confirmed that fuzzy matches are being returned on the Beta Cluster:
Thanks for writing this up, @awight.
Tue, Mar 2
Could this be caused by the styles in rEPOP7b0937c5a865: Use CSS clip-path instead of SVG when supported. not being flipped by CSSJanus?
Nice work on the proof of concept!
Mon, Mar 1
I'm actually going to move this one back into Needs Analysis. AFAICT @awight's approach in rEEVLaca489f139cd: User edit count bucketing is one @Krinkle advised against in T210106#5829916 so we should probably chat about it.
Fri, Feb 26
If my understanding is correct, using time the page was open might include instances where the page is just open in the background during tabbed browsing. This could skew our results and understanding of how long a user is actually viewing the page before they switch languages.
Thu, Feb 25
Yes! Thanks for the ping, @Jdlrobson.
Tue, Feb 23
Mon, Feb 22
Regrettably, this will have to tested by checking data in the event.searchsatisfaction Hive table once the change has gone live – it'll be deployed to the group0 wikis tomorrow – as there's no way to override the sampling code that enables the SearchSatisfaction instrument.
Tue, Feb 16
Excellent! I'd asked @Mholloway if he was looking for a merger just the other day. Great work, everyone!
From an engineering perspective only, this is roughly an S – it's a one-line config change and deployment. The config change is as follows:
I've scheduled the above for deployment.
Mon, Feb 15
Fri, Feb 12
Thu, Feb 11
Wed, Feb 10
Tue, Feb 9
Mon, Feb 8
Thu, Feb 4
Instrument: Clicks to language list in control group
Instrumentation: Time on page prior to switching to a different language
Wed, Feb 3
☝️ Per our discussion yesterday, we'll sign this off now and test the change after we've enabled the A/B test on the Beta Cluster.
Feb 2 2021
Instrumentation: Initial and final language
Instrumentation: Number of new users accessing language switching functionality
Instrumentation: Clicks to new button
AC: Allow logged-in (logged-out users nice to have) to be bucketed into either the new treatment or the control