By default this is linked to the `tests` boolean, which we disabled for Embroider builds in 96674859. We want deprecation-workflow features to be available in production builds, so let's enable it unconditionally.
Until now, we have allowed testing themes in production environments via `/theme-qunit`. This was made possible by hacking the ember-cli build so that it would create the `tests.js` bundle in production. However, this is fundamentally problematic because a number of test-specific things are still optimized out of the Ember build in production mode. It also makes asset compilation significantly slower, and makes it more difficult for us to update our build pipeline (e.g. to introduce Embroider).
This commit removes the ability to run qunit tests in production builds of the JS app when the Embdroider flag is enabled. If a production instance of Discourse exists exclusively for the development of themes (e.g. discourse.theme-creator.io) then they can add `EMBER_ENV: development` to their `app.yml` file. This will build the entire app in development mode, and has a significant performance impact. This must not be used for real production sites.
This commit also refactors many of the request specs into system specs. This means that the tests are guaranteed to have Ember assets built, and is also a better end-to-end test than simply checking for the presence of certain `<script>` tags in the HTML.
Mixins are considered deprecated by Ember, and cannot be applied to modern framework objects. Also, the coupling they introduce can make things very difficult to refactor.
This commit takes the state/logic from BulkTopicSelection and turns it into a helper object. This avoids it polluting any controllers/components its included in.
In future, the entire helper object can be passed down to child components so that they don't need to lookup controllers using the resolver. Those kinds of changes would involve changing some very heavily-overridden templates, so they are not included in this PR. It will likely be done as part of the larger refactor in https://github.com/discourse/discourse/pull/22622.
- Add data-embroider-ignore to all script tags which are not currently being compiled by embroider
- Ensure all remaining script tags are wrapped in `<discourse-chunked-script>` so that Rails will follow any renames which Embroider makes (e.g. when it adds fingerprints to filenames)
Discourse core now builds and runs with Embroider! This commit adds
the Embroider-based build pipeline (`USE_EMBROIDER=1`) and start
testing it on CI.
The new pipeline uses Embroider's compat mode + webpack bundler to
build discourse code, and leave everything else (admin, wizard,
markdown-it, plugins, etc) exactly the same using the existing
Broccoli-based build as external bundles (<script> tags), passed
to the build as `extraPublicTress` (which just means they get
placed in the `/public` folder).
At runtime, these "external" bundles are glued back together with
`loader.js`. Specifically, the external bundles are compiled as
AMD modules (just as they were before) and registered with the
global `loader.js` instance. They expect their `import`s (outside
of whatever is included in the bundle) to be already available in
the `loader.js` runtime registry.
In the classic build, _every_ module gets compiled into AMD and
gets added to the `loader.js` runtime registry. In Embroider,
the goal is to do this as little as possible, to give the bundler
more flexibility to optimize modules, or omit them entirely if it
is confident that the module is unused (i.e. tree-shaking).
Even in the most compatible mode, there are cases where Embroider
is confident enough to omit modules in the runtime `loader.js`
registry (notably, "auto-imported" non-addon NPM packages). So we
have to be mindful of that an manage those dependencies ourselves,
as seen in #22703.
In the longer term, we will look into using modern features (such
as `import()`) to express these inter-dependencies.
This will only be behind a flag for a short period of time while we
perform some final testing. Within the next few weeks, we intend
to enable by default and remove the flag.
---------
Co-authored-by: David Taylor <david@taylorhq.com>
We have the max_mentions_per_chat_message site settings; when a user tries
to mention more users than allowed, no one gets mentioned.
Chat messages may contain code-blocks with strings that look like mentions:
def foo
@bar + @baz
end
The problem is that the parsing code considers these as real mentions and counts
them when checking the limit. This commit fixes the problem.
`badge.save(["name", "description", "badge_type_id"])` api it was testing isn't a thing anymore.
Also: replaces `assert.expect(0)` with more useful assertions
We have a workaround so that currentUser/siteSettings/appEvents work properly on RestModel instances which are created without an owner. This is not ideal, but fixing this properly is not trivial. This commit improves the workaround to be more robust and support all service injections.
Currently, if the review queue has both a flagged post and a flagged chat message, one of the two will have some of the labels of their actions replaced by those of the other. In other words, the labels are getting mixed up. For example, a flagged chat message might show up with an action labelled "Delete post".
This is happening because when using bundles, we are sending along the actions in a separate part of the response, so they can be shared by many reviewables. The bundles then index into this bag of actions by their ID, which is something generic describing the server action, e.g. "agree_and_delete".
The problem here is the same action can have different labels depending on the type of reviewable. Now that the bag of actions contains multiple actions with the same ID, which one is chosen is arbitrary. I.e. it doesn't distinguish based on the type of the reviewable.
This change adds an additional field to the actions, server_action, which now contains what used to be the ID. Meanwhile, the ID has been turned into a concatenation of the reviewable type and the server action, e.g. post-agree_and_delete.
This still provides the upside of denormalizing the actions while allowing for different reviewable types to have different labels and descriptions.
At first I thought I would prepend the reviewable type to the ID, but this doesn't work well because the ID is used on the server-side to determine which actions are possible, and these need to be shared between different reviewables. Hence the introduction of server_action, which now serves that purpose.
I also thought about changing the way that the bundle indexes into the bag of actions, but this is happening through some EmberJS mechanism, so we don't own that code.
This patch adds a new shortcut to allow archiving private messages. When
on a private message page, just type `a` to archive it. Typing `a` on an
already archived message will move it back to inbox.
Chat review queue flags were missing the context message above the actions.
This is probably because the (reasonably complex) logic was somewhat hard-coded to posts. After some investigation I concluded we can reuse this logic with some small amendments.
Previously we were patching ember-cli so that it would split the test bundle into two halves: the helpers, and the tests themselves. This was done so that we could use the helpers for `/theme-qunit` without needing to load all the core tests. This patch has proven problematic to maintain, and will become even harder under Embroider.
This commit removes the patch, so that ember-cli goes back to generating a single `tests.js` bundle. This means that core test definitions will now be included in the bundle when using `/theme-qunit`, and so this commit also updates our test module filter to exclude them from the run. This is the same way that we handle plugin tests on the regular `/tests` route, and is fully supported by qunit.
For now, this keeps `/theme-qunit` working in both development and production environments. However, we are very likely to drop support in production as part of the move to Embroider.
The changes in e1d27400f5 slightly changed the sourcemap paths of our workbox assets. The sourcemaps now have the extension `.prod.map` instead of `prod.js.map`. However, since the version number of workbox didn't change, the directory digest remained the same, and so cached versions of the JS were pointing to the now-nonexistant map files.
This commit introduces a cachebusting constant which we can bump for these kinds of changes in future.
Our Ember build compiles assets into multiple chunks. In the past, we used the output from ember-auto-import-chunks-json-generator to give Rails a map of those chunks. However, that addon is specific to ember-auto-import, and is not compatible with Embroider.
Instead, we can switch to parsing the html files which are output by ember-cli. These are guaranteed to have the correct JS files in the correct place. A <discourse-chunked-script> will allow us to easily identify which chunks belong to which entrypoint.
In future, as we update more entrypoints to be compiled by Embroider/Webpack, we can easily introduce new wrappers.
Previously applied in 2c58d45 and reverted in 24d46fd. This version has been updated for subfolder support.
This will allow us to extend the deprecation period for this-property-fallback beyond Ember 4.x, to give more time for plugin developers to update their templates.
1. Use `this.` instead of `{{action}}` where applicable
2. Use `{{fn}}` instead of `@actionParam` where applicable
3. Use non-`@` versions of class/type/tabindex/aria-controls/aria-expanded
4. Remove `btn` class (it's added automatically to all DButtons)
5. Remove `type="button"` (it's the default)
6. Use `concat-class` helper
When an upload fails and we don't have a specific error, we
show a generic one. But it's a little too generic -- it doesn't
even include the file name.
This commit shows the file name so you at least know which of your
uploads failed.
This tab doesn't really provide anything useful, and can be quite
confusing in some cases. Each plugin is already listed below, and
you can navigate to their settings from there. We want to move away
from the catch-all Plugins category for site settings. Core plugins are
not shown in this list as at 97a812f022.
Our Ember build compiles assets into multiple chunks. In the past, we used the output from `ember-auto-import-chunks-json-generator` to give Rails a map of those chunks. However, that addon is specific to ember-auto-import, and is not compatible with Embroider.
Instead, we can switch to parsing the html files which are output by ember-cli. These are guaranteed to have the correct JS files in the correct place. A `<discourse-chunked-script>` will allow us to easily identify which chunks belong to which entrypoint.
In future, as we update more entrypoints to be compiled by Embroider/Webpack, we can easily introduce new wrappers.
This commit fixes an issue from 2ecc8291e8
where the user sees an ugly plain #hashtag when sending a chat
message. Now, we add a basic placeholder that looks like the
cooked hashtag with a grey square, which is then filled in
once the "sent" message bus event for the message comes back,
and we do decorateCooked on the message to fill in the proper
hashtag details.
When flagging a chat message, and options included both notifying user and notifying staff, the modal was missing the separating text. This was happening because the #staffFlagsAvailable method was based on post flags, and the model for chat flags is slightly different. This fixes that by delegating to the relevant flag target object.
Streaming doesn't work for anonymous users because we scope updates to the current user. Since they can only see cached summaries, we can skip the streaming parameter and update it directly with the ajax response.
Previously, when dragging the timeline scroller, we would position it purely based on the current cursor position. That means that if you started dragging it from anywhere other than the centre, the scroller will 'jump' suddenly to centre itself on the cursor.
This commit measures the offset of your cursor when the drag starts, and maintains it throughout the drag. So for example, if you start dragging at the bottom of the scroller and drag one pixel up, the scroller will only move by 1px.
This should make the UX much smoother, especially on large topics.
When the 'loading slider' navigation indicator is enabled, and a connection is very slow, we `display: none` most of the page and display a spinner. The `still-loading` body class for this was being added in the `afterRender` step in the Ember runloop. This meant that, depending on the order they were scheduled, other `afterRender` jobs may run before it. This caused an issue with topic scroll locations because we would attempt to scroll to an element which was `display: none` at the point its position was calculated.
This commit moves the `still-loading` class manipulations to the `render` step of the runloop, which is technically more correct, and means that anything scheduled in the `afterRender` step is guaranteed to run without the `display: none` CSS.
https://meta.discourse.org/t/276305/29
This commit contains a few improvements:
* Use LinkTo instead of a button with a weird action referencing the
controller to navigate to the filtered settings for a plugin
* Add an AdminPlugin model with tracked properties and use that when
toggling the setting on/off and in the templates
* Make it so the Settings button for a plugin navigates to the correct
site setting category instead of always just going to the generic
"plugins" one if possible
Follow-up to #23199 in which we moved the "delete user" options under the relevant action menu for flagged post. This change does the same, but to queued posts.
This provides a `refresh()` function on Ember's public router service. The feature was added to Ember in v4.1.0, but this polyfill will allow us to start using it straight away under 3.28
DEV: Display fuzzy site setting search results below direct matches
When searching for site settings, in the results under the ALL category
all the fuzzy search results were showing first followed by any direct
matches. This change adjusts that so that fuzzy searches show below
direct matches.
Fuzzy results are now also sorted based on their gap calculation in
ascending order.
It is hard to catch and debug potential bugs related to live updates of
user status (though, we haven't seen many such bugs so far). We have
this `console.warn` statement that should help us to catch one class of
such bugs:
70f1cc5552/app/assets/javascripts/discourse/app/models/user.js (L1384-L1389)
But in tests, we quite often operate with stubs of users that don't have ID,
and this warning create unnecessary noise. This PR disable the warning in
the testing environment.
This could happen after you had already change the separation mode and would cause unexpected bugs.
This PR also adds more tests around using switch buttons with chat.
Reverts e2705df and re-lands #23187 and #23219.
The issue was incorrect order of execution of Rails' `assets:precompile` task in our own precompilation stack.
Co-authored-by: David Taylor <david@taylorhq.com>
Short answer -- the problem is the video thumbnail generator & uploader
code added a couple of months back in f144c64e13.
It was implemented as another Mixin which overrides `this._uppyInstance`
when uploading the video thumbnail after the initial upload is complete,
which means the composer's `this._uppyInstance` value is overridden,
and it loses all of its preprocessors & upload code.
This is generally a problem with the Mixin based architecture that I
used for the Uppy code, which we need to remove at some point and
refacotr.
The most ideal thing to do here would be to convert this video thumbnail
code into an Uppy
[postprocessor](https://uppy.io/docs/uppy/#addpostprocessorfn) plugin,
which runs on each upload after they are complete. I started looking
into this, and the main hurdle here is adding support to tracking the
progress of postprocessors to
[ExtendableUploader](cf42466dea/app/assets/javascripts/discourse/app/mixins/extendable-uploader.js)
so that is out of scope at this time.
The fix here makes it so the ComposerVideoThumbnailUppy code is no
longer a Mixin, but acts more like a normal class, a pattern which
we have used in chat. I also clean up a lot of the thumbnail uploader
code and remove some unnecessary things.
Attempted to add a system spec, but video streaming does not work
in Chrome for Testing at this time, and it is needed for the
onloadedmetadata event.
`Window`'s `resize` event was unreliable. You could shrink down the browser window so that the timeline would disappear but the progress element would not render to replace it.
This commit makes it rely on a media query listener instead so it 1. matches the css 2. fires only when that query result changes (perf win)
By default, the workbox-expiration plugin will not expire cache entries which include a `Vary` header in the response. This means that cached entries can build up until the browser storage quota is hit.
This commit introduces the `ignoreVary: true` option, so that deletion is performed correctly. This will only apply going forward, so this commit also bumps the cache version and deletes the old caches.
Ref https://github.com/GoogleChrome/workbox/issues/2206
This fixes the problem reported in
https://meta.discourse.org/t/custom-status-message-in-front-of-by-header-on-scroll/273320.
This problem can be reproduced with any tooltip created using the DTooltip
component or the createDTooltip function.
The problem happens because the trigger for tooltip on mobile is click, and for tooltip
to disappear the user has to click outside the tooltip. This is the default behavior
of tippy.js – the library we use under the hood.
Note that this PR fixes the problem in topics, but not in chat. I'm going to investigate and
address it in chat in a following PR.
To fix it for tooltips created with the createDTooltip function, I had to make a refactoring.
We had a somewhat not ideal solution there, we were leaking an implementation detail
by passing tippy instances to calling sides, so they could then destroy them. With this fix,
I would have to make it more complex, because now we need to also remove onScrool
handlers, and I would need to leak this implementation detail too. So, I firstly refactored
the current solution in 5a4af05 and then added onScroll handlers in fb4aabe.
When refactoring this, I was running locally some temporarily skipped flaky tests. Turned out
they got a bit outdated, so I fixed them. Note that I'm not unskipping them in this commit,
we'll address them separately later.
This commit adds some system specs to test uploads with
direct to S3 single and multipart uploads via uppy. This
is done with minio as a local S3 replacement. We are doing
this to catch regressions when uppy dependencies need to
be upgraded or we change uppy upload code, since before
this there was no way to know outside manual testing whether
these changes would cause regressions.
Minio's server lifecycle and the installed binaries are managed
by the https://github.com/discourse/minio_runner gem, though the
binaries are already installed on the discourse_test image we run
GitHub CI from.
These tests will only run in CI unless you specifically use the
CI=1 or RUN_S3_SYSTEM_SPECS=1 env vars.
For a history of experimentation here see https://github.com/discourse/discourse/pull/22381
Related PRs:
* https://github.com/discourse/minio_runner/pull/1
* https://github.com/discourse/minio_runner/pull/2
* https://github.com/discourse/minio_runner/pull/3
This PR changes how we track which lists are available for a topic and how we decide which is the active one. The new approach centralizes everything in the service, and exposes functions for adding/removing a list, which each calls via `did-insert/will-destroy` modifiers.
It makes it much easier to track and update state when navigated to another topic or PM, ensuring things get updated correctly.
Separate mobile templates are a pattern we're moving away from. They are not supported by Ember, and make things more difficult to develop/test. The differences between the mobile and desktop templates for `discovery/topics` are very minimal, so they can be easily integrated.
The only feature missing from the main template was the new 'list header controls' UI. This commit introduces that to the main template inside an `mobileView` conditional.
Key changes in behavior, many of which could be considered bug fixes, are:
- Mobile will now include 'redirected reason'
- Mobile will now include shared drafts
- Mobile will now include before-topic-list and after-topic-list Plugin Outlets
- Mobile will now have a `<div class="show-more">` wrapper around the 'new or updated' UI, to match desktop. This does not seem to cause any visual change.
Mobile-specific template overrides of `discovery/topics` will continue to function as before - this should not be a breaking change for any themes/plugins.
Mobile-specific templates for the topic list and topic-list-item remain in place.
Sometimes the fuzzy search would return too many site setting results
making it hard to find what you are searching for. This change still
allows for fuzzy searching but tightens up the criteria for being a
fuzzy match.
One example is searching for 'cheer', a term associated with a plugin,
previously returned ~55 search results. With this change it will return
~13 (Actual numbers depend on how many plugins your instance has).
Another example is searching for 'digest'. Previously returned ~37
results and now will return ~14.
Follow up to: e63e193a0a
See also: https://meta.discourse.org/t/276013
Transpiling to `/raw-templates` is important so that they are detected by the `eager-load-raw-templates` initializer. Prior to 16c6ab86 this wasn't a problem because all connector modules were being eager-loaded. Now that connectors are lazily loaded, it's critical that `eager-load-raw-templates` does the eager loading. This problem doesn't affect themes because they compile raw templates into an iife instead of a `define()` module.
Unfortunately we don't have any way to introduce automated testing for this part of our compilation pipeline. However, discourse-calendar will begin depending on this functionality imminently, so its tests will warn us of future regressions.
This commit introduces a 'shortcut' when rendering PluginOutlets which have no registered connectors. On my machine, this improves rendering performance of empty PluginOutlets by around 30-40% (tested by running tachometer on a `/latest` route with 600 plugin outlets).
* Minor style adjustments
* Removes "all" count because it's redundant to the count on New
* Updates generic class names with -- modifier to follow BEM and help avoid class name collisions
* Hides the toggle when bulk select is enabled (the UI ends up being too busy)
