On Sun, Jun 27, 2010 at 12:06 PM, Thomas Dalton <thomas.dalton(a)gmail.com>wrote;wrote:
On 27 June 2010 19:48, Rob Lanphier
<robla(a)wikimedia.org> wrote:
For example, let's say that there are three
pending revisions in the
queue.
That means there is the latest accepted revision
(we'll call "A1"), and
three pending revisions ("P1", "P2", and "P3"). P3 is the
latest pending
revision, while P1 and P2 are intermediate pending revisions.
The specification says that when viewing the diff between A1 and P3, the
"reject" button is enabled. A more conservative school of thought says
that
the "reject" button shouldn't be
enabled, because its possible that P1
was a
valid revision that was vandalized by P2, and the
only way to tell is to
look at the revision history. However, this should be reasonably rare,
and
the diff remains in the edit history to be
rescued, and can be reapplied
if
need be. A competing problem is that disabling
the "reject" button will
result in the same confusion we're already seeing today.
The guidance for reviewing multiple edits
(
http://en.wikipedia.org/wiki/Wikipedia:Reviewing#Step-by-step_.22how-to.22_…
)
says you have to go through them one-by-one (unless they are all by
the same user), so I suggest eliminating the option of review multiple
edits with a single click, unless they are all by the same user. The
feature should be designed to fit in with the way it is used, after
all. Once you've done that, the issue you raise goes away.
I think it actually gets worse. What should the reject button do in the
case that the reviewer is looking at A1 and P1?
However, I
would suggest a "rollback" or
"undo" button (which does that same as
those buttons always do) rather than a "reject" button - don't
introduce a new term when it does the same thing as an existing
feature with its own name.
The confirmation page that is shown when the user hits "reject" tells the
reviewer that they are about to "undo" one or more revisions. We're not
wedded to the word "reject", but it's pretty clear that reviewers are going
to look around to the counterpart to "accept". There's already an
"undo"
link on these pages, but people still feel that some sort of "reject" or
"decline" is necessary.
Rob