[WikiEN-l] Flagged Protection Revert Etiquette (Re: Renaming "Flagged Protections")

[Rob Lanphier]

On Sat, May 22, 2010 at 12:05 AM, Carcharoth <carcharothwp at googlemail.com>wrote:

> Can you reject with a "let's discuss on the talk page"? What I am
> thinking is that some people use edit summaries to alert other editors
> to a talk page discussion, and if this is not possible with the
> FlaggedRevs system, I would be inclined to accept an edit and then
> revert it and suggest a talk page discussion.
>



At this very moment, there is no "reject" button.  That's one of the last
minute features that we're working on here:
http://flaggedrevs.labs.wikimedia.org/wiki/Wikimedia:Reject_Pending_Revision

There is already a comment entry field beside the "approve" (accept) button
right now, and there's no reason I can see why that same comment entry field
won't be used for the proposed "reject" button.

There is currently a "de-approve" button (which we're renaming "unaccept"),
but that's only for the rare case where someone has accepted a revision, and
then you're unmarking it.



> What I'm asking is whether you need to accept first or not.


Nope, you shouldn't need to.



> I get the
> impression from what you are saying that you can click undo
> straightaway and that automatically accepts the edit and undoes it in
> one step (I would replace the automatic undo summary).



Not quite.  This is what a whiteboard is really handy for  :)  I'm guessing
you may have a slightly incorrect way of thinking about how the feature
works, and that's causing some confusing in cases like this.

It may be helpful to understand how this feature works under the hood to be
able to visualize what's happening.  Each revision has a flag associated
with it (the "accepted" flag) which by default is "false" (unaccepted).
 Approving/accepting an article flips that flag to "true" (accepted).  The
article that gets shown is the latest one with the flag set "true"
(accepted).

The thing that's very confusing for people is that they want to think about
three states for a given revision:  "approved", "rejected", and
"unreviewed".  That's not the way the feature is implemented though.
 "Rejected" and "unreviewed" are indistinguishable at the database level.

There is a distinction that's a fair approximation for "rejected" versus
"unreviewed", which is by answering the question "is there a later accepted
revision than this unaccepted revision I'm looking at?"  If there is, then
the revision is implictly "rejected".  If there isn't, then the revision is
implicitly "unreviewed".  That's how this feature works.  We treat
unaccepted revisions after the latest accepted revision as "pending
revisions".

So, back to your question.  When you click "undo" on a pending revision,
there's no magic accepting going on of the pending revision.  Instead,
you're just putting an accepted revision after it, thus implicitly rejecting
that revision.



> Normally, when
> reverting and adding a custom edit summary, I load the previous page
> version and save that with an edit summary. But I don't think that
> will work here, though maybe it will.
>

Yup, that will still work just fine.



> I suspect that any action by an autoconfirmed user will automatically
> accept something of any actions not yet reviewed. Will those
> autoconfirmed users get a warning that they might unwittingly be
> accepting edits they might not have reviewed?
>

Yup, they do.  There's a banner at the top of the page that tells them
exactly this.

Rob