- Change activate user from admin controller to return json
- Test that it returns json
- Remove unnessary test from log_out spec
This commit was created so that when you activate a user through the api
it returns a json response.
This commit helps improve the discourse_api experience so that we can
check the json response if it was a success or not. This commit also
checks that a 404 is sent instead of a 500 if a bad user_id is passed
in.
* Editing a badge's title would show it as changed in the side even if
you didn't hit save
* Clicking a badge would not scroll to the top
* If there was an error saving a badge there was a missing i18n key
* URLs were using queryParams instead of paths
* User `label` tags for checkboxes for larger click targets
* Saved! text would persist when viewing another badge
* After creating a new badge it would show nothing
* Validation errors were not being properly released to the client
* Query errors were surrounded by an extra array
Changed internals so trust levels are referred to with
TrustLevel[1], TrustLevel[2] etc.
This gives us much better flexibility naming trust levels, these names
are meant to be controlled by various communities.
Upon saving a badge or requesting a badge result preview,
BadgeGranter.contract_checks! will examine the provided badge SQL for
some contractual obligations - namely, the returned columns and use of
trigger parameters.
Saving the badge is wrapped in a transaction to make this easier, by
raising ActiveRecord::Rollback on a detected violation.
On the client, a modal view is added for the badge query sample run
results, named admin-badge-preview.
The preview action is moved up to the route.
The save action, on failure, triggers a 'saveError' action (also in the
route).
The preview action gains a new parameter, 'explain', which will give the
output of an EXPLAIN query for the badge sql, which can be used by forum
admins to estimate the cost of their badge queries.
The preview link is replaced by two links, one which omits (false) and
includes (true) the EXPLAIN query.
The Badge.save() method is amended to propogate errors.
Badge::Trigger gets some utility methods for use in the
BadgeGranter.contract_checks! method.
Additionally, extra checks outside of BadgeGranter.contract_checks! are
added in the preview() method, to cover cases of null granted_at
columns.
An uninitialized variable path is removed in the backfill() method.
TODO - it would be nice to be able to get the actual names of all
columns the provided query returns, so we could give more errors
All flags should end up in one of the three dispositions
- Agree
- Disagree
- Defer
In the administration area, the *active* flags section displays 4 buttons
- Agree (hide post + send PM)
- Disagree
- Defer
- Delete
Clicking "Delete" will open a modal that offer to
- Delete Post & Defer Flags
- Delete Post & Agree with Flags
- Delete Spammer (if available)
When the flag has a list associated, the list will now display 1
response and 1 reply and a "show more..." link if there are more in the
conversation. Replying to the conversation will NOT give a disposition.
Moderators must click the buttons that does that.
If someone clicks one buttons, this will add a default moderator message
from that moderator saying what happened.
The *old* flags section now displays the proper dispositions and is
super duper fast (no more N+9999 queries).
FIX: the old list includes deleted topics
FIX: the lists now properly display the topic states (deleted, closed,
archived, hidden, PM)
FIX: flagging a topic that you've already flagged the first post
Allows for a quick and easy group membership management on the
user-administration page. Uses the select2 UI component to
autosuggest other groups, remove existing ones and lock in automatic
groups.
