We did some testing and saw that making one query per month is
cheaper than querying all chat messages at ones. Note that even
though the export job will be performing one query per month,
the exported messages will be streamed into a single CSV file, so
nothing changes from the user's point of view.
This is extracted from #22390.
This patch introduces a scope to avoid duplication and a new method,
`Chat::Channel.find_by_id_or_slug` to allow finding a channel either by
its id or by its slug (or its category slug).
Prior to this commit we were loading a large number of thread messages without any pagination. This commit attempts to fix this and also improves the following points:
- code sharing between channels and threads:
Attempts to reuse/share the code use in channels for threads. To make it possible part of this code has been extracted in dedicated helpers or has been improved to reduce the duplication needed.
Examples of extracted helpers:
- `stackingContextFix`: the ios hack for rendering bug when momentum scrolling is interrupted
- `scrollListToMessage`, `scrollListToTop`, `scrollListToBottom`: a series of helper to correctly scroll to a specific position in the list of messages
- better general performance of listing messages:
One of the main changes which has been made is to remove the computation of visible message during scroll, it will only happen when needed (update last read for example). This constant recomputation of `message.visible` on intersection observer event while scrolling was consuming a lot of CPU time.
This is extracted from #22390.
This patch simplifies a little how we handle uploads in chat, relying on
ActiveRecord mechanisms instead of calling custom methods.
This also makes `Chat::Message#validate_message` a “real” AR validation,
meaning it will run automatically when `#valid?` is called.
This only moves code around and doesn't change any behavior. This does two things:
1. Extracts the `channel.joined_by?` methods
2. Uses term "members" instead of "participants" for chat members
Turns out making a html4 fragment and then operating on parts of it using html5 fragments is a bad idea. ;)
This seems to fix the issue with occasionally missing GH icons in oneboxes.
Someone who cannot chat is also not able to join chat channels,
so we may not check all the time user.can_chat? && user.can_join_chat_channel?
and just call user.can_join_chat_channel? instead.
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.
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.
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
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.
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
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>
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>
* FEATURE: Sort thread list by unread threads first
This commit changes the thread list to show the threads that
have unread messages at the top of the list sorted by the
last reply date + time, then all other threads sorted by
last reply date + time.
This also fixes some issues by removing the last_reply
relationship on the thread, which did not work for complex
querying scenarios because its order would be discarded.
* FIX: Various fixes for thread list loading
* Use the channel.threadsManager and find the channel first rather
than use activeChannel in the threads manager, otherwise we may
be looking at differenct channels.
* Look at threadsManager directly instead of storing result for threads
list otherwise it can get out of sync because of replace: true in
other places we are loading threads into the store.
* Fix sorting for thread.last_reply, needed a resort.
This small patch registers a new `ActiveModel` type: `array`.
It will split a string on `,` to create a new array. If the value is
already an array, nothing will happen and for all other types, it will
wrap the value in an array.
Here’s an example on an existing contract:
```ruby
attribute :target_usernames
before_validation do
self.target_usernames =
(
if target_usernames.is_a?(String)
target_usernames.split(",")
else
target_usernames
end
)
end
# can be rewritten as:
attribute :target_usernames, :array
```
To export chat messages, go to `/admin/plugins/chat` and click the Create export
button in the _Export chat messages_ section. You'll receive a direct message
when the export is finished.
Currently, this exports all messages from the last 6 months, but not more than
10000 messages.
This exports all chat messages, including messages from private channels and
users' direct conversations. This also exports messages that were deleted.
We had a bug in this code recently, sometimes users saw weird notifications
like:
User mentioned all_mentioned_user_ids in the help chat channel
We fixed that bug in b85d057.
This refactoring is a follow-up to that fix. As that bug showed, it’s quite easy
to introduce a key that may end up being sent to the `NotifyMentioned` job,
which can lead to such weird notifications. This refactoring makes sure that
the `to_notify` hash contains only IDs of users that should be notified about
mentions.
Previously this was defining methods like `UserOption#never?`, `UserOption#all_new?`, `UserOption#dm_and_mentions?`. Now they will be prefixed like `UserOption#chat_header_indicator_never?`
Unfortunately, Discourse's `UserOption` model is not currently autoloaded. That means that modifying it via a `reloadable_patch` will try to apply the changes repeatedly. Normally this doesn't matter since the changes are idempotent. However, introducing ActiveRecord enums is not idempotent - they raise an error if the same enum already exists on the model. This commit adds a check to avoid hitting this 'duplicate definition' error.
Reproduced
```
rails runner 'Rails.application.reloader.reload!'
```
This commit adds a tracking dropdown to each individual thread, similar to topics,
that allows the user to change the notification level for a thread manually. Previously
the user had to reply to a thread to track it and see unread indicators.
Since the user can now manually track threads, the thread index has also been changed
to only show threads that the user is a member of, rather than threads that they had sent
messages in.
Unread indicators also respect the notification level -- Normal level thread tracking
will not show unread indicators in the UI when new messages are sent in the thread.
This commit adds the initial part of thread indicator improvements:
* Show the reply count, last reply date and excerpt,
and the participants of the thread's avatars and
count of additional participants
* Add a participants component for the thread that
can be reused for the list
* Add a query class to get the thread participants
* Live update the thread indicator more consistently
with the last reply and participant details
image image
In subsequent PRs we will cache the participants since
they do not change often, and improve the thread list
further with participants.
This commit also adds a showPresence boolean (default
true) to ChatUserAvatar, since we don't want to show the
online indicator for thread participants.
---------
Co-authored-by: chapoi <charlie@discourse.org>
When a thread is created / a new message is created in the
thread, we want to make sure that the original message user
has a membership for that thread, otherwise they will not
receive unread indicators for messages in the thread.
This patch introduces policy objects to chat services. It allows putting
more complex logic in a dedicated class, which will make services
thinner. It also allows providing a reason why the policy failed.
Some change has been made to the service runner too to use more easily
these new policy objects: when matching a failing policy (or any failing
step actually), the result object is now provided to the block. This
way, instead of having to access the reason why the policy failed by
doing `result["result.policy.policy_name"].reason` inside the block,
this one can be simply written like this:
```ruby
on_failed_policy(:policy_name) { |policy| policy.reason }
```
This commit makes some fundamental changes to how hashtag cooking and
icon generation works in the new experimental hashtag autocomplete mode.
Previously we cooked the appropriate SVG icon with the cooked hashtag,
though this has proved inflexible especially for theming purposes.
Instead, we now cook a data-ID attribute with the hashtag and add a new
span as an icon placeholder. This is replaced on the client side with an
icon (or a square span in the case of categories) on the client side via
the decorateCooked API for posts and chat messages.
This client side logic uses the generated hashtag, category, and channel
CSS classes added in a previous commit.
This is missing changes to the sidebar to use the new generated CSS
classes and also colors and the split square for categories in the
hashtag autocomplete menu -- I will tackle this in a separate PR so it
is clearer.
This moves chat tracking state calculation for channels
and threads into a central Chat::TrackingStateManager service, that
serves a similar purpose to the TopicTrackingState model
in core.
This service calls down to these query classes:
* ThreadUnreadsQuery
* ChannelUnreadsQuery
To get the unread_count and mention_count for the appropriate
channels and threads.
As well as this, this commit refactors the client-side chat
tracking state.
Now, there is a central ChatTrackingStateManager Ember Service
so all tracking is accessible and can be counted from one place,
which can also initialize tracking from an initial payload.
The actual tracking counts are now maintained in a ChatTrackingState
class that is initialized on the `.tracking` property of both channel and
thread objects.
This removes the attributes on UserChatChannelMembership and decoration
of said membership from ChannelFetcher, preferring instead to have an additional
object for tracking in the JSON.
A follow-up to 54b2a85b. That commit didn't fix the issue because the to_notify hash that we return from the notify_edit method isn't used anywhere apart from tests (that's confusing, we're going to fix that soon).
In the past, we create a `chat_mention` records only when we wanted to notify a user about a mention. Since we don't send notifications when a user mentioning himself, we didn't create a `chat_mention` records in those cases.
Now we use `chat_mentions` records in other scenarios too, so when a user is mentioning himself we want to:
1. Create a `chat_mention` record for that mention
2. Do not create a notification for that mention
When the user sends a message in a thread, we want to
create a membership for them in the background (default
to notification level of Watching) so we can track whether
they have read the thread.
Then, for now since we don't have granular message reading/
scrolling in the thread panel, we just update the thread
last_read_message_id for the user to the latest reply in the
thread when they open the thread panel. This at least will
mark the thread as read.
In future PRs we want to show the blue dot indicator in various
places in the UI for unread threads which will also require
some MessageBus functionality.
This takes into account the same issue fixed for channels
in ae3231e140
When setting DISCOURSE_ZEITWERK_EAGER_LOAD=1 to enable
eager loading the previous solution to adding chat_levels
to the core NotificationLevels would break with a module
loading error (c.f. cc2570fce3)
We don't actually _need_ to extend the core class, we can just
make our own for chat, let's do this instead.
This will enable us to begin work on user tracking
state for a thread so we can show thread-specific
unreads and mentions indicators. In this case are following
the core notification_level paradigm rather than the solution
UserChatChannelMembership went with, and eventually we
will want to refactor the other table to match this as well.
Co-authored-by: Joffrey JAFFEUX <j.jaffeux@gmail.com>
After this change, in order to join a chat channel, a user needs to be in a group with at least “Reply” permission for the category. If the user only has “See” permission, they are able to preview the channel, but not join it or send messages. The auto-join function also follows this new restriction.
---------
Co-authored-by: Martin Brennan <martin@discourse.org>
This commit adds an initial thread list UI. There are several limitations
with this that will be addressed in future PRs:
* There is no MessageBus reactivity, so e.g. if someone edits the original
message of the thread it will not be reflected in the list. However if
the thread title is updated the original message indicator will be updated.
* There is no unread functionality for threads in the list, if new messages
come into the thread there is no indicator in the UI.
* There is no unread indicator on the actual button to open the thread list.
* No pagination.
In saying that, this is the functionality so far:
* We show a list of the 50 threads that the user has most recently participated
in (i.e. sent a message) for the channel in descending order.
* Each thread we show a rich excerpt, the title, and the user who is the OM creator.
* The title is editable by staff and by the OM creator.
* Thread indicators show a title. We also replace emojis in the titles.
* Thread list works in the drawer/mobile.
When making the list of users to notify we set `all_mentioned_user_ids` key on the `to_notify` Hash.
This hash will be passed around until the actual moment where we send the notifications:
```ruby
identifier_text =
case identifier_type
when :here_mentions
"@here"
when :global_mentions
"@all"
when :direct_mentions
""
else
"@#{identifier_type}"
end
```
As not found `all_mentioned_user_ids` would end up being sent as `@all_mentioned_user_ids` which is obviously incorrect.
This commit is a direct fix to the issue and will remove the key as soon as we have used it sooner up in the chain.
This bug was reproducible when doing this sequence of events:
- create a message with a direct mention: `@bob hi`
- edit this message into a global mention `@all hi`
Every replies creates a thread, even when threading is disabled. This is how we ensure we can go back and forth. However, a message bus event should only be published when threading is enabled, otherwise frontend will attempt to display a thread which is not possible when disabled.
This fixes a silent background 404 when doing a reply in a direct message channel or a non threading enabled category channel.
What is the problem?
Previously, this was the query used to move change messages into another
channel.
```
INSERT INTO chat_messages(
chat_channel_id, user_id, last_editor_id, message, cooked, cooked_version, created_at, updated_at
)
SELECT :destination_channel_id,
user_id,
last_editor_id,
message,
cooked,
cooked_version,
CLOCK_TIMESTAMP(),
CLOCK_TIMESTAMP()
FROM chat_messages
WHERE id IN (:message_ids)
RETURNING id
```
The problem is that this incorrectly assumes that the insertion will be based on the order of `message_ids`. However, that
is not the case as PostgreSQL provides no such guarantee. Instead we need to explicitly order the messages to ensure
the right order of insertion.
This problem was discovered by a flaky test which exposed the non-guarantee order of insertion.
We need to create and update `chat_mentions` records for messages earlier. They should be created or updated before we call `Chat::Publisher.publish_new!` `Chat::Publisher.publish_edit!` to send the message to message bus subscribers).
This logic is covered with tests in `message_creator_spec.rb`, `message_updater_spec.rb`, `notifier_spec.rb` and `notify_mentioned_spec.rb`.
See the commits history for steps of refactoring.
This commit implements all the necessary logic to create thread seamlessly. For this it relies on the same logic used for messages and generates a `staged-id`(using the format: `staged-thread-CHANNEL_ID-MESSAGE_ID` which is used to re-conciliate state client sides once the thread has been persisted on the backend.
Part of this change the client side is now always using real thread and channel objects instead of sometimes relying on a flat `threadId` or `channelId`.
This PR also brings three UX changes:
- thread starts from top
- number of buttons on message actions is dependent of the width of the enclosing container
- <kbd>shift + ArrowUp</kbd> will reply to the last message
This was reverted in 38cebd3ed5.
The issue was that I was using Discourse.redis.delete_prefixed
which does a slow redis KEYS lookup, which is not advised in
production. This commit removes that, and also ensures the periodical
thread count update only happens if threading is enabled.
I changed to use a redis INCR/DECR for reply count
cache. This avoids a round trip to redis to GET the current
count, and also avoids multi-process issues, where
if there's two processes trying to increment at the
same time, they may both receive the same value, add one
to it, then both write the same value back.
Then, it's only n+1 instead of n+2.
This also prevents almost all chat scheduled jobs from
running if chat is disabled, the only one remaining is
the message retention job.
This commit moves the category channel creation out
of the Chat::Api::Channel controller and into a
dedicated CreateCategoryChannel service. A follow up
commit will move the DM channel creation out of
the old DirectMessageChannelCreator service.
Also includes a new on_model_errors helper
for chat service class usage, that collects model
validation errors to present in a nice way.
---------
Co-authored-by: Loïc Guitaut <loic@discourse.org>
Andrei Prigorshnev
Loïc Guitaut
Joffrey JAFFEUX
Jan Cernik
Jarek Radosz
Martin Brennan
David Taylor
Alan Guo Xiang Tan
Daniel Waterworth