Last modified: 2013-05-14 18:20:46 UTC
Created attachment 11626 [details] Screenshot of a page pending review right after undo I have just undid https://pl.wikipedia.org/w/index.php?title=Kurhaus_Publishing&diff=34150229&oldid=34148206 creating https://pl.wikipedia.org/w/index.php?title=Kurhaus_Publishing&diff=34209359&oldid=34150229 and after reload I got "The stable version was checked on 9 January 2013. There is 1 pending change awaiting review." A screenshot is attached. A previous version was manually checked: 20:59, 9 January 2013 John Belushi (Talk | contribs | block) reviewed a version of Kurhaus Publishing (changes reviewed) (revision: 20:59, 9 January 2013) https://pl.wikipedia.org/w/index.php?title=Specjalna%3ARejestr&type=review&user=John+Belushi&page=Kurhaus+Publishing&year=&month=-1&tagfilter= My permissions: Administrators, Autoconfirmed users, Check users, Abuse filter editors, Editors, Users From https://pl.wikipedia.org/wiki/Special:UserGroupRights: Editors (list of members) Edit semi-protected pages (autoconfirmed) Have one's own edits automatically marked as "checked" (autoreview) Mark revisions as being "checked" (review) Quickly rollback the edits of the last user who edited a particular page (rollback) View recent changes patrol marks (patrolmarks) View the list of unreviewed pages (unreviewedpages) I had reviewed the change manually (and it worked).
More information: https://pl.wikipedia.org/w/index.php?title=Kurhaus_Publishing&diff=34146931&oldid=31508069 was a recreation of the page and it was not reviewed at all (older edits have been restored from deleted revisions). Therefore the manual review was the first, initial one (i-review) and this revision was undone. I would expect that the page returns to the initial (before i-review) state and that all edits go live immediately and are not subject to review. If the previous edit would not have been manually i-approved, the undo would work as is.
Since the top X (1 in this case) edits where reverted, it's equivalent to reverting to the previous revision. Since that revision was not reviewed, the new edit was treated as needing review (since it was treated as "based off" that unreviewed revision). I think it could be changed so that using ?undo= always treats the "base" as the current. It could decouple the "equivalent" revision as another concept. That way, if either the base or equivalent revision is reviewed, the new edit can be reviewed. This can be done with an extra hidden input on the edit form.
Related URL: https://gerrit.wikimedia.org/r/60909 (Gerrit Change Ibd36b60df6cff0d04e035b81855b2789b22a9240)
> I think it could be changed so that using ?undo= always treats the "base" as > the current. It could decouple the "equivalent" revision as another concept. > That way, if either the base or equivalent revision is reviewed, the new edit > can be reviewed. This can be done with an extra hidden input on the edit > form. Not sure I understand what's written above but I'll try the gerrit patch :)