559102 - User reviews are associated with latest add-on version, even if it's in the sandbox

Status: RESOLVED FIXED

(addons.mozilla.org Graveyard :: Public Pages, defect, P3)

Description

[Jorge Villalobos [:jorgev] (he/him)]

According to a report filed by an author on the amo-editors list, he recently deleted a version of his add-on that was in the sandbox, only to find several user reviews disappearing.

There are two problems here IMO. First, it's not clear for users or authors that these reviews are version-dependent. Unless we make this clear for them I think it would make more sense to remove this dependency and associate reviews with add-ons instead of versions. Secondly, the most likely target version for a review is the one currently visible to the public, not the latest version, which could be in the sandbox or could be a beta version. If we want to continue associating reviews to versions, by default they should be set to the latest public version, and perhaps users should have the option of choosing which version they're reviewing.

Comment 1

[Michael Buckley]

I don't see why we even allow them to delete versions of an extension.  Why can't they just remove it from public, if they really have to, and upload a new version?

Comment 2

[Russell Gilbert]

Hi Michael, I'm the add-on developer that originally reported the problem.

In this particular case, the version I was deleting was still in the sandbox.  It had been there a couple of days, and I needed to release a new version, so I removed it from the sandbox and added the newer one -- even though I knew it would go to the end of the line for reviewing.

The problem was that when I removed the sandboxed version, the six new reviews I had received since putting that version into the sandbox were deleted also -- even though reviews don't seem to be connected to a particular version, and that particular version hadn't been made public yet anyway.

To make things worse, one of the reviews that was deleted was a change to an existing review, where one of my users got things working ok, and changed a 1 star review to a 5 star review. Since the change got deleted, it went back to the 1 star review, and I now have no way of reporting it via the web site, since it's already listed as being reported (I've sent two additional emails about this actually -- I'd really appreciated if someone could take care of this -- please just delete the latest two reviews, from Chef and Don.O).

Thanks!
Russell

Comment 3

[Russell Gilbert]

Sorry, I just realized that something I said in the previous comment might not have been clear.

In the second paragraph, what I meant was that the version in the sandbox had been made invalid (in this case, because of a change Facebook made to their site), so there was no use in waiting for it to be reviewed. The new version I added to the end of the queue fixed the problem, so it was best to just remove the sandboxed version before it was reviewed and start over.

Thanks,
Russell

Comment 4

[Michael Buckley]

Reviews are deleted, just might take a few minutes before that shows because of caching.

I still ask, what did you achieve by deleting the version?  So that this needs to be allowed at all?

Comment 5

[Russell Gilbert]

Thanks for deleting the reviews.

The version I had waiting for review had become unnecessary, and I needed to replace it with a new version to be reviewed.

You asked earlier why I didn't just remove it from the public and upload a new version.  But the version I deleted had never been made public -- it was still waiting to be reviewed. The only option I saw was to delete that version and upload another one. Is there some other way I could have done it?

Comment 6

[Jorge Villalobos [:jorgev] (he/him)]

Michael, you should look at bug 544486, that's about not deleting versions or files permanently.
Russell, you can just upload a new version without deleting any of your current versions. We prefer that approach.
Let's try to stay on topic, please.

Comment 7

[Russell Gilbert]

Ok, but in this case, the version I deleted was still awaiting review. Since I no longer wanted that version to be reviewed, how should I have stopped the review process for that version without deleting it?

What I'm trying to say is that I had a version that was public (let's call it version 1), and another that was awaiting review (version 2). I decided I no longer wanted version 2 to be reviewed, and wanted to upload version 3 and have it reviewed instead. It was when I deleted version 2 that the six comments were removed with it.

Comment 8

[Jorge Villalobos [:jorgev] (he/him)]

Uploading version 3 would have made version 3 take version 2's place in the review queue, and version 2 would have never been reviewed.

Comment 9

[Russell Gilbert]

Ok, that's good to know. I'll do that in the future.

I think the bug, as originally written, is still valid though. I don't think user reviews should be tied to particular versions since they're not presented that way, and certainly not to versions that have not yet been released to the public.

Comment 10

[Jeff Balogh (:jbalogh)]

Fixed in http://github.com/jbalogh/zamboni/commit/6d4a5f3.  The review is associated with the latest current version (the version we show on the details page).