Use local-storage service, prototyped here https://github.com/LevelbossMike/local-storage-service, to manage local storage usage in an octane way. Does not write to local storage in tests by default and is easy to stub out.
Copy passed hash before manipulating it.
Assigning to the same hash object will break href-to
because in certain scenarios href-to-helper will
not create a new object that gets passed to
`fsm-with-optional`-hrefTo method.
This is problematic for optional route-params, and lead
to a situation where links to peered services would
create the wrong url for their href-attribute.
We need to explicitly tell the UI to not show the bucket-list
when we are displaying imported services. If we make
this depend on the data we will sometimes not show
it due to data-loader caching.
Working with a peer model as a relationship is much
easier than to workaround a non-relationship in
imported services. This is currently only relevant
for imported-services where we know the peer
in advance.
Add a helper that knows how to format past dates
in a smart way. When less than a week ago we will
use relative date strings - for dates older than a week
we will use a friendly human-readable format.
This matches best practices we want to adhere to
based on what Terraform did for date-formatting.
To measure the available space of an element when it
should take up the "rest" of the page. This matches
what `ListCollection` is doing internally but makes
the mechanism available in a composable component.
We need a component abstraction that encapsulates
creating the dynamic tabs based on peering-type.
We create a `PeerTab`-abstraction that behaves like
the data-structure the tab-nav expects to achieve this
effect.
* Update empty-state copy throughout app
Update empty-states throughout the app to only include mentions of ACLs if the user has ACLs enabled.
* Update peers empty state copy
Flip the empty state copy logic for peers. Small typo fixes on other empty states.
* Update Node empty state with docs
* Update intentions empty state
Make ACL copy dependent on if acls are enabled.
* Update Nodes empty state learn copy
* Fix binding rule copy key
* Use postcss instead of ember-cli-sass
This will make it possible to work with tailwindcss.
* configure postcss to compile sass
* add "sub-app" css into app/styles tree
* pin node@14 via volta
Only used by people that use volta
* Install tailwind and autoprefixer
* Create tailwind config
* Use tailwind via postcss
* Fix: tailwind changes current styling
When adding tailwind to the bottom of app.scss we apparently
change the way the application looks. We will import
it first to make sure we don't change the current styling
of the application right now.
* Automatic import of HDS colors in tailwind
* Install @hashicorp/design-system-components
* install add-on
* setup postcss scss pipeline to include tokens css
* import add-on css
* Install ember-auto-import v2
HDS depends on v2 of ember-auto-import so we need to upgrade.
* Upgrade ember-cli-yadda
v0.6.0 of ember-cli-yadda adds configuration for webpack.
This configuration is incompatible with webpack v5
which ember-auto-import v2 is using.
We need to upgrade ember-cli-yadda to the latest
version that fixes this incompatability with auto-import v2
* Install ember-flight-icons
HDS components are using the addon internally.
* Document HDS usage in engineering docs
* Upgrade ember-cli-api-double
* fix new linting errors
* Wrap service names on show and instance routes
Moves the trailing type/kind/actions to the second row of the header
no matter what length the service name is. Wraps service name text.
* Change grid format of AppView globally
* Add tooltips to the last element of breadcrumbs
* v3.20.2...v3.24.0
* Fix handle undefined outlet in route component
* Don't use template helper for optional modal.open
Using the optional-helper here will trigger a computation
in the same runloop error. This is because we are setting
the `modal`-property when the `<Ref>` component gets
rendered which will update the `this.modal`-property which
will then recompute the `optional`-helper leading to this
error.
Instead we will create an action that will call the `open`-method
on the modal when it is defined. This gets rid of the double
computation error as we will not access the modal property
twice in the same runloop when `modal` is getting set.
* Fix - fn needs to be passed function tab-nav
We create functions in the component file instead
so that fn-helper stops complaining about the
need to pass a function.
* Update ember-exam to 6.1 version
"Makes it compatible" with ember-qunit v5
* scheduleOnce setMaxHeight paged-collection
We need to schedule to get around double-computation error.
* Fix - model.data is removed from ember-data
This has been private API all along - we need to
work around the removal.
Reference: https://github.com/emberjs/data/pull/7338/files#diff-9a8746fc5c86fd57e6122f00fef3155f76f0f3003a24b53fb7c4621d95dcd9bfL1310
* Fix `propContains` instead of `deepEqual` policy
Recent model.data works differently than iterating attributes.
We use `propContains` instead of `deepEqual`. We are only
interested in the properties we assert against and match
the previous behavior with this change.
* Fix `propContains` instead of `deepEqual` token
* Better handling single-records repo test-helper
`model.data` has been removed we need to handle proxies and
model instances differently.
* Fix remaining repository tests with propContains
We don't want to match entire objects - we don't care
about properties we haven't defined in the assertion.
* Don't use template helper for optional modal.open
Using a template helper will give us a recomputation error -
we work around it by creating an explicit action on
the component instead.
* Await `I $verb the $pageObject object` step
* Fix no more customization ember-can
No need to customize, the helper handles destruction
fine on its own.
* Fix - don't pass `optional` functions to fn
We will declare the functions on the component instead.
This gives us the same behavior but no error from
`fn`, which expects a function to be passed.
* Fix - handle `undefined` state on validate modifier
StateChart can yield out an undefined `state` we need
to handle that in the validate modifier
* Fix linting errors tests directory
* Warn / turn off new ember linting issues
We will tackle them one by one and don't want to
autofix issues that could be dangerous to auto-fix.
* Auto-fix linting issues
* More linting configuration
* Fix remaining linting issues
* Fix linting issues new files after rebase
* ui: Remove ember-cli-uglify config now we are using terser (#14574)
Co-authored-by: John Cowen <johncowen@users.noreply.github.com>
* ui: Add AWS Lambda as potential external source
* Update mocks to expect `lambda` as an external-source meta
* Additional mock to make sure all our mocks only contain whats possible
* Fix await page-object properties properly
* Remove "sleep"-hack show-routing test
* Fix - change serviceInstacesSelected test order
We need to change the order of test as the tab-nav
tab-elements are actually clickable and will be triggered
via the "see" step.
Because we are now properly awaiting these promises
we need to change the test order and check if the
service-instances tab is selected first, as the following
steps will actually click the respective tabs during
testing and thus change what is selected
on the tab-nav.
* Update ember-cli-page-object to latest
* Refactor remaining `moduleFor`-tests
`moduleFor*` will be removed from ember-qunit v5
* Upgrade ember-qunit to v5
* Update how we use ember-sinon-qunit
With ember-qunit v5 we need to use ember-sinon-qunit differently.
* Fix submit-blank test
We can't click on disabled buttons with new test-helpers.
We need to adapt the test accordingly.
* Make sure we await fill-in with form yaml step
We need to await `fill-in`. This changes the reducer
function in the step to create a proper await
chain.
* Fix show-routing test
We need to await a tick before visiting again.
* Remove redundant `wait one tick`-step
* remove unneeded "next Tick" promise from form step
* Increase timeout show-routing feature
* Comment on pause hack for show-routing test
* Include nspace when surfacing peer in bucket-list
Whenever we display a peer and we are not on OSS we will surface
the namespace as well. The rest of the ui logic of the bucket list
has not changed.
* Display bucket-list after instance-count service-list
* ui: Make DataWriter also deal with more standard shaped errors
* ui: Add error states/messages for peering establishment (#14026)
* ui: Add error states/messages for peering establishment
* Inspect state instead of asking about existence of error
* ui: Make peered intentions read-only
* Replace "" to undefined for SourcePeer so its the same as PeerName
* Fixup copypasta
* Ensure tests run with no peers
* add ember-cli-deprecation-workflow
* Add deprecation workflow configuration
This will silence all deprecations by default reducing noise in the test
output significantly.
We can tackle deprecations now one by one but won't have to deal with
very verbose console logs anymore.
* Only show instances- and tags-tab peered services
* Adapt show-with-slashes test to peering changes
Tests always have the peering feature turned on and the default service
we load from the mock-api will be peered. This is why the topology
view of the service.show page will not be accessible in the updated
test it will show the instances instead. This change does not change
what the test is actually testing so just putting changing to the now
different url is fine.
* don't show partition / peer at the same time in bucket-list
* use bucket-list in intentions table
* add bucket-list tests
* Simplify bucket list - match old behavior
Refactor the bucket-list component to be easier to grok and match
how the old template based approach worked. I.e. do not surface
partition or namespace when it matches the passed nspace or partition
property.
* Update docs for bucket-list
* fix linting
* ui: Add a modal.opened property for inspecting whether the modal is open
* merge isOpen setting into the exiting event handler
* Revert to multiple listeners, plus comment to explain
* Wrap close in an afterRender