Why this change?
The test being changed in question has been flaky on our CI. However, we
are unable to view the screenshot of why it failed because
ActionDispatch will only take a screenshot of the default session upon
failure. At the same time, taking screenshot of all sessions
automatically upon failure is not possible via the official Capybara or
Rails APIs at the moment. Therefore, we're changing this system test to
avoid using two custom session and instead have the main assertion use
the default session such that any failures will provide us with a
screenshot.
This commit fixes two issues with the thread list:
1. All threads were being shown regardless of whether the user had
a membership in the thread. This was happening because the list
and the channel share the same thread store, so if the channel
had OMs with threads we would load them and they showed in the list.
2. Threads created by the user from a staged thread would double up.
This is because the _cache in the channel threadsManager would use
the staged thread ID even after we'd replaced the object's ID with
the actual thread from the DB. The answer to this is to remove and
re-add the thread to the local cache with the actual ID.
* DEV: Fix and re-enable chat flakys
The early return in JS was added to prevent an error
from channel being null, and it's better to use known
users for the message fabrications in the specs.
* DEV: Use travel_to in drawer spec for thread tracking
Sometimes in the system test the datetime that is last
viewed for the channel for the user and the datetime for
the last message created_at is only microseconds of difference,
and we do not provide that level of fidelity in the MessageBus
serializer, so unreadThreadsCountSinceLastViewed is not
accurate.
Better to just utilize travel_to and move forward 1 minute in
time before sending the new message to easily differentiate.
When we have subscriptions for new messages in a channel,
we also have special handling for messages in a thread. For
cases like DM channels where threads are made in the background
but not used in the UI, this is causing JS errors because we
are trying to fetch the thread but it returns 404.
We only want to do things with messages in threads if the
channel actually has threading enabled.
We need a nice way to only return some hashtag data
sources based on various site settings. This commit
adds an enabled? method that every hashtag data source
must implement. If this returns false the data source
will not be used at all for hashtag lookups or search.
On tablets like iPad where we allow channel and thread to be on the same screen, it was not possible to resize the panels due to code being thought for mouse events. This commit should now correctly allow for this.
The "resizer" has also been made larger to simplify touching.
No test as it's hard to test on iPad and dragging events are also complex.
On iOS we have a hack to prevent the viewport to move when focusing an input, however this code was targeting the textarea node through a global selector which is working fine on iOS as we only show one composer at a time but was failing on iPad as we show both channel and thread on the same screen. As a result `document.querySelector(".chat-composer__input")` was always targeting the first textarea on the screen which was the channel's composer, making it impossible to focus the thread's one.
This commit makes it so that when the user has unread threads
for a channel we show a blue dot in the sidebar (or channel index
for mobile/drawer).
This blue dot is slightly different from the channel unread messages:
1. It will only show if the new thread messages were created since
the user last viewed the channel
2. It will be cleared when the user views the channel, but the threads
are still considered unread because we want the user to click into
the thread list to view them
This necessitates a change to the current user serializer to also
include the unread thread overview, which is all unread threads
across all channels and their last reply date + time.
Why this change?
The test being changed in question has been flaky on our CI. However, we
are unable to view the screenshot of why it failed because
ActionDispatch will only take a screenshot of the default session upon
failure. At the same time, taking screenshot of all sessions
automatically upon failure is not possible via the official Capybara or
Rails APIs at the moment. Therefore, we're changing this system test to
avoid using two custom session and instead have the main assertion use
the default session such that any failures will provide us with a
screenshot.
Prior to this change you might end up in a loop where removing a channel would redirect you to this channel and as we auto-follow opened direct message channels, you could never unfollow this last direct message channel.
Followup to d7ef7b9c03,
this adds a spec to test the case where old threads are
still unread for the user and should show at the top regardless
of pagination, and fixes some issues/makes some slight refactors.
This commit attempts to fix an issue where we are ending
up with bad created_at date formats for last messages, which
is breaking the DM sort order and sometimes causing DM channels
to fall off the list, or show "Invalid date" on mobile.
I have not been able to consistently reproduce these issues
locally, however the serialzier for the channels index uses
MultiJSON.dump() and the Chat::Publisher uses .to_json, both of
which format created_at differently for messages.
The former is `2023-07-05T06:53:25.977Z` (iso8601).
The latter is `2023-07-14 03:59:22 UTC` (.to_s default).
Since we are doing comparison and sorting of these dates on the UI
we need consistent formatting for the JS Date parsers (and moment)
to deal with.
If the issue still occurs after this we can investigate further.
Prior to this commit a long press on the image of a chat message would trigger both the actions menu and the contextual menu. This commit ensures we only show the contextual menu in this case.
No test as it's a quite complex behavior to reproduce (would need android for example).
This was causing this event to cause other touch events down the road. For example click a reaction above the composer when the message action was opened could cause the composer to gain focus after the reaction was made.
It could only occur on message created by the user itself and deleted while the user was looking at the channel.
It more generally fix the trash service which was not correctly setting the author of the delete.
`SiteSetting.enable_public_channels` allows site admin to decide if public channels are available at all. There's no distinction between admins or not as we expect admins to create private category channels if they want to limit usage.
Not sure how this was even working previously, since it's trying
to press the reply button on a thread original message, which doesn't
work, you need to click the indicator to open the thread.
Why this change?
The following test is flaky on our CI:
```
1) Navigation when sidebar is configured as the navigation menu when re-opening full page chat after navigating to a channel opens full page chat on correct channel
Failure/Error: measurement = Benchmark.measure { example.run }
expected "/" to equal "/chat/c/random-9/17"
```
The theory here is that system tests is running too fast that we're not
giving the href for the chat header icon a chance to update before
clicking on it. Therefore, we're adding an additional assertion to
assert that the link has the right href before clicking on it.
Initial migration and changes to models as well as
changing the following services to update last_message_id:
* Chat::MessageCreator
* Chat::RestoreMessage
* Chat::TrashMessage
The data migration will set the `last_message_id` for all existing
threads and channels in the database.
When we query the thread list as well as the channel,
we look at the last message ID for the following:
* Channel - Sorting DM channels, and channel metadata for the list of channels
* Thread - Last reply details for thread indicators and thread list
It is now safe to render the message excerpt as HTML since
it is no longer using text_entities: true in the server
PrettyText.excerpt call when creating the message excerpt
from the cooked HTML.
This will fix the issue of things like mentions showing
HTML code instead of the actual mention when replying,
and cannot be used to inject improper HTML like style tags
via XSS.
This commit makes sure we don't load all data into memory when doing CSV exports.
The most important change here made to the recently introduced export of chat
messages (3ea31f4). We were loading all data into memory in the first version, with
this commit it's not the case anymore.
Speaking of old exports. Some of them already use find_each, and it worked as
expected, without loading all data into memory. And it will proceed working as
expected after this commit.
In general, I made sure this change didn't break other CSV exports, first manually, and
then by writing system specs for them. Sadly, I haven't managed yet to make those
specs stable, they work fine locally, but flaky in GitHub actions, so I've disabled them
for now.
I'll be making more changes to the CSV exports code soon, those system specs will be
very helpful. I'll be running them locally, and I hope I'll manage to make them stable
while doing that work.
* FEATURE: Inline topic summary. Cached version accessible to everyone.
Anons and non-members of the `custom_summarization_allowed_groups_map` groups can see cached summaries for any accessible topic. After the first 12 hours and if the posts to summarize have changed, allowed users clicking on the button will automatically re-generate it.
* Ensure chat summaries work and prevent model hallucinations when there are no messages.
Browser capabilities are inherently unconnected to the lifecycle of our app. Making them formally available outside of the service means that they can safely be used in non-app-linked functions without needing risky hacks like `helperContext()` or `discourse-common/lib/get-owner`.
One example of where the old hacks were problematic is the `translateModKey()` utility function. This is called in the root of the `discourse/components/modal/keyboard-shortcuts-help` es6 module. If anything (e.g. a theme/plugin) caused that es6 module to be `require()`d before the application was booted, a fatal error would occur.
Following this commit, `translateModKey()` can safely import and access `capabilities` without needing to worry about the app lifecycle.
The only potential downside to this approach is that the capabilities data now persists across tests. If any tests need to 'stub' capabilities, they will need to revert their changes at the end of the test (e.g. by using Sinon to stub a property).
This commit also updates some legacy references from `capabilities:main` to `service:capabilities`.
This implementation will need more work in the future. For simplification of tracking and other events (new thread, delete/restore OM...) we used the threads from `threadsManager` which makes pagination more complicated as we already have some results when we start.
Note this commit also simplify `Collection` to only have one `load` method which can be called repeatedly.
Why this change?
The specs are flaky on CI and we've unable to figure out why so we've
decided to skip them only on CI for now. The tests are still ran in our
internal build so we still have some protection in place.
Trying to fix two issues:
1. Sometimes the publish_new! event for update_thread_original_message
finishes running on the UI before the one for thread_created, in this
case we just want to do nothing because thread_created will fetch the
new thread along with its preview from the server if needed
2. Sometimes the thread GET and /read events were erroring because
last_reply on the thread was nil, this was potentially occuring because
the thread_created event was coming through to the UI before the rest
of MessageCreator was done, so we just move that after the big update
to set thread_id for the new and existing messages in the reply
chain
Why is this change being made?
We've decided that the previous "community" section should look more
like a primary section that holds the most important navigation links
for the site and the word "community" doesn't quite fit that
description. Therefore, we've made the decision to drop the
section heading for the community section.
As part of removing the section heading, the following changes are made
as well:
1. Button to customize the section has been moved to the "footer" of the
"More..." section when `navigation_menu` site setting is set to `sidebar`.
When `navigation_menu` is set to `header dropdown`, a button to customize
the section is shown inline.
2. The section will no longer be collapsable.
3. The title of the section is no longer customisable as it is no longer
displayed. As a technical note, we have not dropped any previous
customisations of the section's title previously in case we have to
bring back the header in the future.
4. The new topic button that was previously present in the header has
been removed alongside the header. Admins can add a custom section
link to the `/new-topic` route if there would like to make it easier for
users to create a new topic in the sidebar.
The `message_bus_channels` given to `MessageBus.last_ids(*message_bus_channels)` is not ordered, as a result the expectation of the tests could fail, this test ensures we check the contain of the input instead of content+order.
This commit also standardize the naming pattern of modals: `<Chat::Modal::FooBar />` and changes css class accordingly.
Co-authored-by: David Taylor <david@taylorhq.com>
When the loading slider is enabled, the rendering of `application.hbs` is slightly delayed compared to the old 'spinner' strategy. This means that if a route tried to render a dialog during its `model()` hook, the dialog wrapper element would not be present and an error would occur.
This commit detects that situation and delays rendering the error until the next runloop iteration. If the element is still not found, we print a useful error to the console.
In the long term, we should ideally convert the dialog service to use a pure-ember rendering strategy instead of leaning on a11y-dialog. But for now, this workaround should resolve the problems identified by the chat system specs.
It's way more common to have presence enabled than disabled, so we should have been making it the default from start.
This commit also changes the namespace of `<ChatUserAvatar />` into `<Chat::UserAvatar />` and refactors tests.
Sadly this function is one of the very hard to test codepaths of the app. We could in the future attempt to extract the content of the function to unit-test it.
When a user sends their first message in a thread we
automatically track the thread in the backend, but we
don't reflect this in the UI until the user re-opens
the thread. This commit fixes that by showing the new
tracking level in the UI.
Chat drawer was using the `DiscourseURL` hook `afterRouteComplete`. This hook suffer from a very poor implementation which makes it very unreliable:
```javascript
if (typeof opts.afterRouteComplete === "function") {
schedule("afterRender", opts.afterRouteComplete);
}
```
This commit attempts to return the promise from `handleURL` to directly use it and have a very reliable after transition hook.
In previous changes we prevented creating a channel to also make users follow the channel. We were forcing recipients to follow the channel on message sent but this was not including the creator of the message itself.
This commit fixes it and also write an end-to-end system spec to cover these cases. The message creator service is currently being rewritten and should correctly test and ensure this logic is present.
This commit also makes changes on the frontend to instantly follow a DM when you open it, this change prevents a green dot to appear for a split second when you send a message in a channel you were previously not following. Only recipients will see the green dot.
Since we create threads in the background regardless of whether
threading is enabled for a channel, we get the unexpected behaviour
of everyone having a lot of unread threads when threading is enabled
for the channel.
To counteract this, when the admin enables threads for a channel
we can just run a high priority background job to mark all threads
as read in the channel for all users, so they are essentially
starting from a clean slate.
Followup to 802fb3b194
We should not hide the replies count if there is only 1 participant
for a thread, because this makes it look like the last reply is the
only reply.
This introduces a PLATFORM_KEY_MODIFIER const that
can be used both client and server side, to determine
whether we should be using the Meta or Ctrl key based
on whether the user is on Windows/Linux or Mac.
Why this change?
Before this commit, there is a chance that we will transition the user
to a different route if the chat thread component has been destroyed
prior to the request for fetching messasges in a chat thread returning.
This commit makes it such that we simply ignore the request if the chat
thread component has been destroyed.
We believe this is the cause of the flaky system tests in plugins/chat/spec/system/navigation_spec.rb
which we've been seeing on CI.
Why this change?
`Faker::Lorem.paragraph` generates a differrent length of string
every time. When a string happens to be long, it can change the UI
across system test runs making it harder to reason about our system
tests across multiple runs since the state is never really consistent.
We will just generate a paragraph with a fixed length going forward so
that the UI remains consistent. This should make certain tests which
relies on the UI being in a certain state to become less flaky.
Why this change?
This change ensures that we scroll to the top of the message when
hovering over a message to ensure that the message actions container
that appears on hover is not hidden in the chat drawer when the content
of the chat message is long.
This commit includes several fixes and improvements to thread
original message handling:
1. When a thread's original message is deleted, the thread no longer
counts as unread for a user
2. When a thread original message is deleted and the user is looking
at the thread list, it will be removed from the list
3. When a thread original message is restored and the user is looking
at the thread list, it will be added back to the list if it was
previously loaded
In specific conditions (generally a small drawer, with a long message) it is possible to have the message’s actions menu to be displayed hover the drawer's header.
This is particularly hard to fix correctly using popper due to our positioning which is slightly at the limit of the container.
The proposed fix targets mostly the specs by ensuring the messages actions will be hidden before attempting to click any header's button.
Why this change?
In CI, we know we're clicking a link to a chat channel's threads list.
However, the threads list is not loaded and we want to add more
assertions here to try and figure out why. By asserting for the current
URL, we will at least know that the transition to the URL is successful.
- Presence needs to be explicitly set on the component now
- We were not checking and testing correctly the presence of the unread indicator in the menu
This commit replaces two existing screens:
- draft
- channel selection modal
Main features compared to existing solutions
- features are now combined, meaning you can for example create multi users DM
- it will show users with chat disabled
- it shows unread state
- hopefully a better look/feel
- lots of small details and fixes...
Other noticeable fixes
- starting a DM with a user, even from the user card and clicking <kbd>Chat</kbd> will not show a green dot for the target user (or even the channel) until a message is actually sent
- it should almost never do a full page reload anymore
---------
Co-authored-by: Martin Brennan <mjrbrennan@gmail.com>
Co-authored-by: Jordan Vidrine <30537603+jordanvidrine@users.noreply.github.com>
Co-authored-by: chapoi <101828855+chapoi@users.noreply.github.com>
Co-authored-by: Mark VanLandingham <markvanlan@gmail.com>
Without this fix, the following error is raised:
```
ActiveRecord::StatementInvalid:
PG::SyntaxError: ERROR: syntax error at or near ")"
LINE 4: WHERE thread_id IN ()
```
Why this change?
Chat system tests that opens the message actions on mobile have been
flaky on our CI. Those system test usually fails when the message
actions do not show up as expected causing subsequent actions to fail.
In the case of the `Reply to message - channel - mobile when the message has an existing thread replies to the existing thread`
system test, failure screenshot shows that we ended up navigating to the
thread instead of opening the message actions button. To understand why
this happens, we first need to understand that by default Capybara clicks
on the centre of an element. Also, we need to note that the HTML structure of
a chat message is like so:
```
<div class="chat-message-container">
<div class="chat-message">
<div class="chat-message-avatar" />
<div class="chat-message-content" />
<div class="chat-message-thread-indicator" />
</div>
</div>
```
Since `PageObjects::Pages::ChatChannel#expand_message_actions_mobile`
attempts to click on the `.chat-message-contaier`, there is a
possibility that the center of that element is the
`.chat-message-thread-indicator` element which would explain why we
navigated to the thread list instead of opening up the message actions.
This is possible because the content of the original chat message as
well as the message excerpt in the thread is randomly generated where the
length of the message and how the text wraps on mobile can affect the
height of the `.chat-message-content` element as thus its position in
the `.chat-message-container` element. In most cases, the middle of the
`.chat-message-container` happens to be the `.chat-message-content`
which is why this test "flakes" sometimes.
What is the solution?
Instead of clicking on the `.chat-message-container`, we be more
specific and click on the `.chat-message-content` element instead.
* UX: make timestamp font size smaller
* UX: participants use copy instead of avatar
* FIX: Move thread participant count into i18n
---------
Co-authored-by: Martin Brennan <martin@discourse.org>
* DEV: Fix flaky thread nav spec
When we transitioned from the chat thread panel under some conditions
the request for the thread would come back and realise the component
was destroyed, which was trying to do a transition to the channel
itself.
Now we check for the previous route here too and transition to the
correct route.
* DEV: Fix chat transcript spec relying on animation
The on-animation-end modifier is not reliable in system specs
because it fires instantly (we have disabled capybara animations)
so the showCopySuccess boolean can be mutated back to false straight
away.
Better to have a separate boolean tracked with a data-attr that we
can reliably inspect in the system spec.
Why this change?
By ensuring the reset happens in an `ensure` code block, we ensure that
the code will always be run even if code fails or an error is raised.
This helps to prevent leaking custom network condition states and
improves the stability of our system tests.
- Inline mentions on posts
- Inline mentions on chat messages
- The user autocomplete for the composer
- The user autocomplete for chat
- The chat section of the sidebar
This fixes a longstanding TODO to move the contents of the
UpdateUserCountsForChannels job to the ensure_consistency!
method of Chat::Channel, which runs every 15 mins as part of
periodical updates.
This commit also addresses the performance issue of the original,
where we would fetch all channels and do an individual query to
get the count and update the count of each one. Now we do it all
in one query, and only publish the changed channels to the UI.
Followup to 3f1024de76
The ActiveModel::Types.register(:array) call for chat was
called too late in the Zeitwerk load order in production,
causing this error:
> `lookup': Unknown type :array (ArgumentError)
> raise ArgumentError, "Unknown type #{symbol.inspect}"
> ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
We need to load the type and register it manually before the rest
of the chat files are loaded via the engine and Zeitwerk.
This will be used when we move the channel creation for DMs
to happen when we first send a message in a DM channel to avoid
a double-request. For now we can just have a new API endpoint
for creating this that the existing frontend code can use,
that uses the new service pattern.
This also uses the new policy pattern for services where the policy
can be defined in a class so a more dynamic reason for the policy
failing can be sent to the controller.
Co-authored-by: Loïc Guitaut <loic@discourse.org>
This has been flagged by our internal system as well and has been
failing on CI. Skip this for now to improve the stability of our system
test runs while we figure out why it is flaky.
Alan Guo Xiang Tan
Martin Brennan
Isaac Janzen
Jan Cernik
Joffrey JAFFEUX
chapoi
Joe
Andrei Prigorshnev
Roman Rizzi
David Taylor
Discourse Translator Bot
Jarek Radosz
Mark VanLandingham
Loïc Guitaut