Should we try to fix existing edits too? This could be tricky as we don't have the recording ID of the created recording anywhere. We could find all recordings with no edit in their history that is: add standalone, add track, edit medium and see if the recording is in use. If so, we someone how try and determine which release it came from?

It'd certainly be nice if we could fix them but it sounds pretty tough; and maybe not worth our while getting such a fuzzy match right?

You'd be pretty much relying on a matching title right? Can you even be sure which artist the "new recording" is/was under; unless you assume that the current release's state represents the "new recording" that was being changed in the earlier edits?

Yea, it'll be difficult, because add medium actually also has this problem...

It would be nice if the UI could show that new recordings have created with this edit.
Maybe by appending "(new)" next to the recording: "Recording title (new)"

This is an important information when evaluating a new release that is added.

Re try fixing existing edits, that would be nice, yes.
As a general note, I'm noticing that we store really too few information in edit data, and this is generating a lot of issues...

As a general note, I'm noticing that we store really too few information in edit data, and this is generating a lot of issues...

Sadly this has always been the case, and I don't know what to do about it. If we try and change it now we have to introduce a new version to most edits and the code gets horribly complicated. We can't upgrade most edits to add missing data though, because we don't have a view of the data at that point in time. http://acidcycles.wordpress.com/2011/07/28/versioning-multiple-versions-of-data-in-relational-databases/ is me trying to explain how I want to do things in BookBrainz (which is a combination of a Git model and a star schema, I think), but I don't know how well it will really scale. If anyone has ideas on how we can get out of this problem I'd love to talk about and invest some time in working on ideas too...

The review actually fixed something entirely different: http://codereview.musicbrainz.org/r/1530/

What's the status of this ticket? Based on the MBS-5241 work I think it should be possible to fix this completely, but I'd like to know what's been done and hasn't already.

Part of my changes for MBS-3514 will at least mark these as "this recording will be added when the edit applies" – i.e., they're marked as new. They do appear to show up in edit histories for the recording at least, which looks like what Oliver's changes were.

Is this sufficient?