I’ve just merged in the Review Queue that has been in development for the last few months. We’ve been running it on our servers for a while and it’s passed through the hands of several developers so it’s in a good place to start testing with those Discourse users who follow the tests-passed branch.
Here are the notes from the commit:
This is a large refactor that merges all types of reviewable items in core Discourse (Users, Queued Posts and Flags) into one consolidated review queue.
New Features
One convenient place to review all content
Reviewable items are scored behind the scenes and the highest priority items are shown first. Staff members and higher trust level users produce higher priority items.
You can filter the review queue by status, priority, reviewable type, category and user
The API for reviewables returns the list of actions that can be performed on an item, which means the front end application can be simpler and not duplicate all the logic.
Data structures are built to enable non-staff level moderation of items in the near future
Software Design Notes
The Reviewable model uses single table inheritance (STI), and each type of reviewable must declare a subclass, serializer and ember component. Three types currently exist: ReviewableQueuedPost, ReviewableUser and ReviewableFlaggedPost. Developers are encouraged to look at how other reviewable items are implemented and build their own.
Post Actions (such as flags) must now be created using the PostActionCreator service object. It has a much friendlier API now and returns a result object which will return the associated reviewable.
Backwards Compatibility
REST API is maintained with the old endpoints, however deprecation warnings are logged. Third party code that is pointed at the old APIs should be updated to the new review queue API. All of the Ember.js/front end code for the old user approval / flagging interface / queued posts is removed.
In a future release we will remove the old APIs and backwards compatibility.
Please use this topic for feedback and bug reports. I’ll try to address things as quickly as possible!
After upgrading from Discourse 2.3.0 beta 5 to beta 6 I now have 14 new notifications in the new /review page, all regarding users that need to be approved.
Those users are either user accounts we rejected weeks ago or banned users, though.
If I clikc on the “Reject” button I get an error message “Sorry, an error has occurred” (see picture below).
TLDR version, I have 14 notifications about already rejected/banned users and I cannot find a way to get rid of them
So the error popping up is almost certainly that the users have posts so they can’t be deleted. I think I realise the bug in my migration now, where I was creating reviewables for users who were not approved. I think perhaps banning someone is setting them as unapproved, so they showed up again.
This fixes the missing translations. Good catch @featheredtoast because it made me realise those components were admin only at the same time, which I also fixed:
I thought I’d seen this complaint elsewhere, but I can’t find it. I just did a g+ import and there are 1300 users in the approval queue with entries like this:
From this it’s impossible to tell why the user needs approval, and worse, the only way to find out seems to be to select the username and do a search. Is there a reason not to have the username link to their profile? Or, better, provide something about the user to know what to do about it?
And also, I presume that “Reject” rejects the user and does Something Bad. How does one clear this and say that the user is OK?
Help! How do I fix this?
My inclination is to just remove them all from the approval queue. But I don’t know how to do that because in the user record I see:
approved: true,
approved_by_id: -1,
approved_at: Fri, 29 Mar 2019 18:26:49 UTC +00:00,
So I guess this approval must be in some other table?
@pfaffman can you help me track this down? You said you did an import, but was it before you ran database migrations? Was must_approve_users set to true?
I updated from the latest a few weeks ago to just now, site is not loading (the Oops message) and shows this in the logs a lot:
Failed to handle exception in exception app middleware : PG::UndefinedTable: ERROR: relation "queued_posts" does not exist
LINE 8: WHERE a.attrelid = '"queued_posts"'::regclass
^
: SELECT a.attname, format_type(a.atttypid, a.atttypmod),
pg_get_expr(d.adbin, d.adrelid), a.attnotnull, a.atttypid, a.atttypmod,
c.collname, col_description(a.attrelid, a.attnum) AS comment
FROM pg_attribute a
LEFT JOIN pg_attrdef d ON a.attrelid = d.adrelid AND a.attnum = d.adnum
LEFT JOIN pg_type t ON a.atttypid = t.oid
LEFT JOIN pg_collation c ON a.attcollation = c.oid AND a.attcollation <> t.typcollation
WHERE a.attrelid = '"queued_posts"'::regclass
AND a.attnum > 0 AND NOT a.attisdropped
ORDER BY a.attnum
There are some gamebreaking problems with this as it stands now, ill run through them here.
When i click on a flagged comment, it does not jump directly to the comment in question in the thread. I have to use the search feature to locate the comment that got flagged. This is very stressful on mobile, and tedious overall.
When a post gets deleted by the system, i lose the ability to go to the thread in question. I prefer to read the thread as it is, and not through the flag menu.
This also made things even more complicated when i as unable to mark it as “Not Spam”, and got prompted with this error:
The way we solved this was to ignore the flag, then locate the deleted thread and un-delete it manually. I dont know if this manual workflow was possible from the beginning, but having the ability to go to the thread in question would certainly be helpful regardless.
I dont know if this is intentional or not, but when a flag is “ignored”, i thought that was only supposed to affect that one person, leaving the flag to be death with by someone else. Please correct me if im wrong. Currently, if someone ignores a flag, it gets ignored for everyone. If this is intentional, then look past this point, however, if its ment to only ignore for you, then thats not working properly.
That being said, i love this update as a whole. Just a few minor things to patch up!
It’s a bit tricky to explain, but you’ll want the error that happened right when you clicked “Not Spam”. You might be able to find it by timestamp. When you’re viewing it, you should see the path similar to /review/(some number)/perform/not_spam.json with an error message.
Alright i have to go, but i’ll try to recreate the scenario once i get home with a new account. If i cant figure it out then i’ll see if Bart can look into it.
