Author: TomH
[Added to the original trac issue at 11.39am, Tuesday, 26th February 2013]

No it's not - that user just has an empty changeset which is hidden in his edit list.

Author: ppawel[at]fastmail.fm
[Added to the original trac issue at 8.30am, Wednesday, 27th February 2013]

Replying to [comment:1 TomH]:

No it's not - that user just has an empty changeset which is hidden in his edit list.

Wouldn't that be the definition of "broken" for this functionality? I mean, the counter shows one thing, the list another without any explanation... Closing as "invalid" does not seem like the best solution to this - IMHO at the very least this issue should be kept on file for reference since it is a bug.

Speaking of a solution for this - I guess some code would be needed around edits counter cache (if we are to stick with this implementation), e.g.

http://blog.douglasfshearer.com/post/17495285851/custom-counter-cache-with-conditions

But that seems to issue additional select on every update - not sure if this is feasible (although if there is a btree index on user_id in the changesets table then it should be pretty quick?).

Author: TomH
[Added to the original trac issue at 8.31am, Wednesday, 27th February 2013]

There's already a bug for that, so we don't need a second.

Author: TomH
[Added to the original trac issue at 8.32am, Wednesday, 27th February 2013]

I would also argue that having the counter hide some of the changesets would clearly be wrong - personally I would just stop hiding them in the list.

Author: ppawel[at]fastmail.fm
[Added to the original trac issue at 9.14am, Wednesday, 27th February 2013]

Replying to [comment:3 TomH]:

There's already a bug for that, so we don't need a second.

Can you link to that bug? I was only able to find #4785 which is closed as well.

Author: TomH
[Added to the original trac issue at 9.17am, Wednesday, 27th February 2013]

Don't know which one is was offhand but I know it came up in the last few weeks.

Yes - the one you found is it, and as I explained it is not realistically fixable.