Why this change?
We have been debugging flaky system tests and noticed in https://github.com/discourse/discourse/actions/runs/7911902047/job/21596791343?pr=25690
that ActiveRecord connection checkout timeouts are encountered because
the Capybara server thread is processing requests even after
`Capybara.reset_session!` and ActiveRecord's `teardown_fixtures` have already been call.
The theory here is that an inflight request can still hit the Capybara
server even after `Capybara.reset_session!` has been called and end up
eating up an ActiveRecord connection for too long and also messing with
the database outside of a transaction.
What does this change do?
This change adds a `BlockRequestsMiddleware` middleware in the test
environment which is enabled to reject all incoming requests at the end
of each system test and before `Capybara.reset_session!` is called. At
the start of each RSpec test, the middleware is disabled again.
Why this change?
The tests are consistently flaky and failing with the following error:
```
CapybaraTimeoutExtension::CapybaraTimedOut:
This spec passed, but capybara waited for the full wait duration (10s) at least once. This will slow down the test suite. Beware of negating the result of selenium's RSpec matchers.
```
Why this change?
On CI, we have been seeing flaky system tests because ActiveRecord is
unable to checkout a connection. This patch is meant to help us debug
which thread is not returning the connection to the queue.
This would allow a theme component (or an API call) to reset the bump
date of a topic to a given post's created_at date.
I picked `post_id` as the parameter here because it provides a bit of
extra protection against accidentally resetting the bump date to a date
that doesn't make sense.
I attempted to add a test for related topics, as they are untested. This was a massive challenge as related topics are only rendered after
- a user has stopped typing
- there is a timeout after typing
Between system tests, acceptance tests, and stubbing requests, this was difficult. So here is the pr san test.
Why this change?
Backfilling can take a long time on a site with many users. As a
result, the `DistriburedMutex` will warn in the logs when the block takes
more than 60 seconds to complete. We can increase the lock validity but
the method is currently only called from a job which has a
`cluster_concurrency` set to `1`. Therefore, it is actually not
necessary for us to hold a `DistributedMutex` here.
What does this change do?
1. Removes the distributed mutex and adds a comment about the risk of
calling the `SidebarSiteSettingsBackfiller#backfill!` method.a
2. Delete old sidebar category section links in batches for all users
to avoid a single long running query.
In certain cases, chat channels may have empty slugs, it happens when:
1. The `slug_generation_method` setting is set to `None`
2. `slug_generation_method` is set to `ASCII` and a channel with
a Unicode name and an empty slug is created (in this case, the code
that creates channels tries to generate a slug and fallbacks to an empty slug)
At the moment, we have a unique index on the `chat_channels.slug` column
which leads to errors when creating several channels with empty slugs
(Discourse is able to create one such channel, but when trying to create
the second one fails because of the unique constraint). This PR fixes that
by adding a `where` condition to the index. Slugs still have to be unique,
but now many channels may have empty slugs.
This fix is similar to the one we made to the category slugs – 7ba914f1e1.
When we launched the new illegal flag type, there were a few problems with the translations:
The translation for the message in the e-mail was missing and in the review queue, the message read: "Is this it's illegal?"
In this PR the missing translation key has been added. For the review queue there was a coupling of the name rendering to whether the flag is of "custom" type, but this is also used for deciding whether we render the textbox for additional details. I think these two things should not be coupled together. For now I have instead hard-coded the existing "custom" types when formatting the name. We can potentially improve this later.
This commit was created with a combination of the ember-native-class-codemod and manual cleanup.
User-status-related functionality was previously encapsulated in its own `User.reopen` call, which is essentially an 'inline mixin'. This commit refactors it into a utility class, with an instance accessible on `User#statusManager`
The service `Chat::CreateMessage` will now accept `context_post_ids` and `context_topic_id` as params. These values represent the topic which might be visible when sending a message (for now, this is only possible when using the drawer).
The `DiscourseEvent` `chat_message_created` will now have the following signature:
```ruby
on(:chat_message_created) do | message, channel, user, meta|
p meta[:context][:post_ids]
end
```
Fixes an issue where private topics that are quoted have an incorrectly formatted url when using a subfolder install.
This update returns a relative url that includes the base_path rather than a combination of base_url + base_path.
Channels can include emojis in front of the channel title which causes problems when sorting.
Using the channel slug is a more reliable way to sort and avoid these kind of issues.
`create topic allowed groups` site setting controls permission to create a new topic.
If a user cannot create a topic, a reply as a new topic button should not be visible as well.
The only addition is PosterSerializer is this:
discourse/app/serializers/poster_serializer.rb
Line 4 in 4a5616f
Which allows themes to check the posters' primary group!
'please refresh, or you may experience unexpected behavior' sounds quite threatening to me. 'please refresh to to keep things working smoothly' conveys the same information in a more friendly way
