On Mon, Jun 28, 2010 at 12:46 AM, Rob Lanphier <robla(a)wikimedia.org> wrote:
"Unaccept" seems suitably rare that I think
we should consider a
confirmation screen which shows the effect of unaccepting (i.e. a diff
between the latest accepted revision and the penultimate accepted revision).
Does that seem like a reasonable enough failsafe to keep this from being
used unintentionally? This seems beneficial even in the case where the
reviewer knew they were hitting "unaccept".
I had the impression that "unaccept" does not add a new revision to
the page, it simply removes the db entry that the revision in question
was "accepted". Is that wrong?
There are two reasons why that ''should'' be the behavior:
(1) If I go back and "unaccept" a revision from two days in the past
that was mistakenly accepted, I should not have to edit the page again
to restore the previous content.
(2) "Accept" does not add a revision to the page, and "unaccept"
should only undo what "accept" does. That is, the sequences "accept =>
unaccept " and "unaccept => accept" should be able to be iterated as
many times as desired, leaving everything in exactly the same state
apart from log entries.
- Carl