When serializing the `body_changes` in the `PostRevisionSerializer`, we create two diffs: one for the `cooked` and another one for the `raw` version of the post.
Inside `DiscourseDiff`, we generate both `html` and `markdown` diffs when we only need the `html` diffs for the `cooked` version of the post and the `markdown` diff for the `raw` version of the post.
This solves the issue repored in https://meta.discourse.org/t/server-error-accessing-topic-revisions-on-a-specific-topic/339185 where some revisions would return 500 because of a `ArgumentError : Attributes per element limit exceeded` exception when trying to generate the `html` diff on a very large `raw`.
This will ensure AI generated titles don't appear as out of range in the UI and also allow users to set longer titles. The limit in DB was already 100 so it's just a simple frontend change.
We were using a complex logic to make it change size based on scroll position but this was imperfect and not visually pleasing. Also the title had been made a button which was causing the ellipsis to not work correctly, and I would prefer to not mix page knowledge (thread) with title component so I made this click logic directly in the chat-thread component.
---------
Co-authored-by: Jordan Vidrine <jordan@jordanvidrine.com>
- Use `requestAnimationFrame` when transitioning from `ready` -> `loading`. The previous `next()` implementation was unreliable, particularly in Safari, and would cause the loading slider to jump backwards instead of forwards
- Double the minimum transition time to 200ms. This avoids the rolling average being skewed too much by routes which load quickly without network access.
This commit ensures that tracked properties added to the post model are correctly synced when using `post.updateFromPost`.
It also introduces a plugin API to allow plugins to register new tracked properties in the post model without needing to modify the class.
Animating the background-color property like this is not compositable for the browser, which means the animation is not smooth, and can contribute to the Cumulative Layout Shift web vital.
For now, we're removing this, and may consider re-introducing an alternative version in future based on user feedback.
The hierarchical search for categories is composed of several complex
nested queries. This change ensures that the secured categories are
filtered out as soon as possible to ensure that the default limit of 5
categories is reached.
Without this fix, the search can return less than 5 categories if any
of the first 5 categories cannot be displayed due to permissions.
Non-admin/moderator users can bulk select items in new/unread, but not in
latest/top/hot. This commit ensures that when the user can no longer
bulk select items in a list, the bulk select checkboxes in the topic list
rows are hidden.
Using CTEs and DISTINCT ON to:
- Pre-filter active users with correct preferences
- Get only first unread message per channel
- Eliminate redundant joins and message scanning
This reduces the query execution time by limiting message scanning and joins to only relevant users and messages.
Internal ref t/142836 & t/139517
A lot of the data of fields is decided at insertion time and is not dynamic afterwards, this commit attempts to solve this problem by making the fk-field-data a component with getters on the all the properties we need. It allows for example to implement a dynamic @disabled without having to pass @disabled everywhere. Generally speaking this solution limits props-drilling.
@format has received the same treatment than @disabled.
In our CI env, sometimes, we see problems regarding getting a connection
from the pool. As it’s currently set to 1, increasing it a bit should
fix that kind of issues.
This commit will hide the channel when the side panel is present and the width of the viewport is less than 1000px. This is especially useful when you want to focus reading a thread on a small screen.
This change only impacts desktops.
In some cases in CI env, it seems the AR connection isn’t available and
the `ensure` block is executed. It’s calling `#verify!` on the
connection, so it can fail sometimes. This is probably why
`#clear_active_connections!` was failing too sometimes.
Here, we just check the connection is present before clearing the
connections.
