It would be useful to be able to filter changes made by new users as it would make catching spam users easier and quickly review changes by genuine new users.
My proposition for the (default) definition of a new user: A new user is a user with less than 20 edits or (yes "or", not "and") less than 7 days since first login.
Maybe a minor difference, but I would change 7 days since first login to 7 days since first edit.
Oh yes, good idea.
Commits pushed to master at https://github.com/mozilla/kuma https://github.com/mozilla/kuma/commit/21e6c629327805d0b1051646e3d2b7ce8e041324 Fix bug 812728: Rev dash - Add filter for changes from new users https://github.com/mozilla/kuma/commit/1d9b56adce627a835ec7a8f7b4be0aea77706064 Merge pull request #932 from Elchi3/dash-newusers-812728 Fix bug 812728: Rev dash - Add filter for changes from new users
This is merged and pushed to production, but the query is very slow, as predicted by :jsocol. (https://github.com/mozilla/kuma/pull/932#discussion_r3414061)
Reopening as there's currently a stack overflow issue.
Commits pushed to master at https://github.com/mozilla/kuma https://github.com/mozilla/kuma/commit/2c279f802c559e90d4bab6f17010befb94bc972b bug 812728 - waffle new user filter https://github.com/mozilla/kuma/commit/e41f3e828d7562f7be8ce5d4510fb69563cb1bce Merge pull request #934 from groovecoder/waffle-dash-new-user-filter-812728 bug 812728 - waffle new user filter
So, what's slow here? Is it getting the new users with the raw query or the creator_id IN(x,y,z, ...) filter (which contains a large amount of ids probably)? If the former, can we cache the raw query then or write a task to get the new users every x minutes only? Or a whole different approach? What would you suggest?
I suspect the issue is with the first query that select ALL the "new users" by looking at ALL revisions ever created. But with it breaking, it's hard to be sure.  https://github.com/mozilla/kuma/pull/932/files#L1R168
I was wrong - it wasn't slow it was broken.
Hm, would love to see bug 665084 fixed here (a more real, local database to test against to for these kind of features would be so cool!). Will see if I manage to limit the user lookup to not to look into ALL revisions.
The vagrant install instructions include a step to download and import a sanitized production data dump: https://github.com/mozilla/kuma/blob/master/docs/installation-vagrant.rst#whats-next
Where does this bug stand? Did it ever get fixed?
Not fixed. Unknown level of effort to do the queries.