--- Comment #43 from Risker <> ---
(In reply to Dan Garry from comment #39)
> Chris's tests were successful. Still, we can afford to take our time with
> this to make sure that it all works correctly in production. Here's the
> rollout plan for this:
> Week 1:
> Week 2: wikiquotes, wikiversity, wikibooks, wikispecies, wikivoyage,
> wiktionary, wikinews, wikisource
> Week 3: commons, meta, anything else I've forgotten except wikipedias
> Week 4: wikipedias
> We'll keep a week between deployments to see if there are any unexpected
> issues, and if there are then we'll delay future deployments. The oversight
> log will remain accessible after the migration so we can verify that
> everything went fine. I'm familiar with the operation of suppression and
> have global suppression by virtue of my staff rights, so I'll be available
> to help if anything goes wrong.
> I'll write on here when I know what date this will begin.

Good plan, Dan.  If I may suggest: 

Week 4: all Wikipedias except enwiki 

Week 5: Enwiki - See note below

Suggested QA reviews:
>10 oversights - all
10>50 oversights - 10 reviews
50>100 oversights -20%
100+ oversights -10 to 15%
Enwiki - 5 to 10%

Suggest enlisting several enwiki oversighters to participate in the QA sample. 
Enwiki represents over 80% of oversighted edits, and if something is going to
be weird, it will probably be here; when I look back at the logs, I see a lot
of really strange actions. 

I've done up a google doc[1] stratifying the proposed schedule, with suggested
number of QA checks, based on the results Reedy got in Comment 16.


You are receiving this mail because:
You are on the CC list for the bug.
Wikibugs-l mailing list

Reply via email to