This commit makes the new AWS sources feature only available in beta.
Note that the RTKQ hooks related to AWS sources are called in several
places outside of the AWS Target step (a prefetch on the Image Output
step and useQuery hook on the review step) but have not been hidden
behind beta flags - this should not present any problems and will make
exposing this feature in stable much easier when the time comes.
This commit adds the ability to specify AWS targets using the sources
service on insights.
This is the first commit to the codebase that makes use of the new RTK
Query endpoints, so I will provide a bit of additional context here:
The sources are obtained by calling the `useGetAWSSourcesQuery()` hook.
This hook can be called in any component where information about the
sources is needed.
A few tricks are used to make the user experience as responsive as
possible.
The `prefetch()` hook provided by RTK Query is called when the user
clicks on the AWS button on the image output step. This triggers the
initial request for the sources, which will then (hopefully) be ready by the
time the user clicks to the next step (the AWS target environment step)
where they are needed.
Because we anticipate a common user workflow to involve using the Create
image wizard in one browser tab and the sources service in another tab,
sources are also refetched every time the source dropdown is opened.
This means that if a user adds a source while in the middle of using the
wizard, they will be able to see it in the wizard's sources dropdown
without refreshing their browser.
Finally, because of the `Recreate image` feature, the
`useGetAWSSourcesQuery` hook also needs to be called on the review step.
Require const declarations for variables that are never reassigned after
being declared. If a variable is never reassigned, using the const
declaration is better. const declaration tells readers, “this variable
is never reassigned,” reducing cognitive load and improving
maintainability.
Due to merging the openshift and chrome plugin systems there's
additional requirements on the format of the module name.
Camelcase is the default, so the values can just be removed from the
config.
Fixes#911. Fixes#912.
This updates the File system step. The changes are:
- changed the toggle group for selecting the mode of partitioning to radio select and updated information for each radio button as mentioned in #911
- added information to the "automatic" option of the file system configuration as mentioned in #912
- updated information for the "manual" option of the configuration as mentioned in mocks
All instances of '3rd party repositories' (and its variations) have been
changed to 'Custom repositories'. This is in better agreement with the
content sources app on consoledot, which generally is called either
'Custom Repositories' or 'Repositories'. Additionally, it is possible to
add and manage Red Hat repositories via content sources, which makes
'3rd party' confusing.
Adds support for 3rd party repositories using the Red Hat Insights
Repositories app on console.redhat.com.
The packages step has been refactored heavily to reduce the bug surface
area and improve its reusability (it is now used in two Wizard steps).
New features related to the Repositories app are currently only exposed
in stage. Because stage and production are quite divergent (they have
different steps, for instance) there are separate test suites for the
production and stage versions of the Wizard. When these features are
moved into production, the two test suites can be merged into one.
'ami' was deprecated a while ago, and is kept around as an alias of
'aws'. Same with 'vhd' and 'azure', and 'vhd' and 'gcp'.
This maintains support for the old aliases where necessary.
The Content Sources service can now be used for handling package info and
search. The api call to this service requires a list of repo urls. The
`repos.json` file tracks the supported repositories for each distro.
This is then mapped into a url array for whichever distro we are
searching for packages. The response body is similar to what existed in
image-builder but the key `package_name` needs to be mapped to just
`name`. Also, there is no pagination in Content Sources for `/rpms/names`
so we can currently only fetch a limit of 20 packages.
The insights platform is moving to react router v6, meaning router
contexts can no longer be nested. As a result all applications will end
up sharing the same `BrowserRouter`.
Switch to using the release (`/` or `/beta/`) as a basename for the
BrowserRouter, and offload the full path (`/insights/$appname`) to the
individual routes. This makes it easier to drop the BrowserRouter in
image builder for the platform one in future.
This commit modernizes the Redux store to bring it in line with the
current recommended best practices/patterns. It is possible because of a
recent previous commit that added Redux Toolkit as a dependency.
For detailed information on what modern Redux entails, see the Redux docs:
https://redux.js.org/introduction/getting-started#learn-modern-redux-livestream
Practically speaking, this means a huge reduction in boilerplate code.
Maintaining and extending the code will be much easier. All Redux logic
is now encapsulated by 'slices'. Reducers are defined in the slices, and
action creators and action types are generated automatically.
Redux Toolkit includes Immer, which greatly simplifies writing reducer logic
much simpler - state updates in Redux must be immutable, but working
with Javascript objects in an immutable fashion is clumsy, requiring
gratuitious use of the spread ... operator. With Immer, the state can be
updated as if mutable updates were allowed. Reducer logic has been
changed to take advantage of this in this commit.
This commit also removes a large amount of unused code. Fingers crossed
that nothing breaks. The tests do pass, afterall...
One other minor change... composesGet was renamed to fetchComposes and
composeGetStatus was renamed to fetchComposeStatus. This is in line with
the Redux documentation examples.
The Details step is renamed to Name Image and moved to the last step
before Review. Also, the validation is removed since all utf-8
characters are supported for image names.
Validation is not enforced on the image output step
when selecting and then deselecting an image type. This
is because the only validation which is set is `Required`.
However, after deselecting an image type the input becomes
a key-value object with all the values set to false. This
commit adds a custom validator to check that at least one
value in the object is set to true. Fixes#575
Resolves#466. The default behavior for the DDF Wizard is to either advance to
the next step or submit the form when the ENTER is pressed.
This is inappropriate for complex forms that contain many different
types of elements that are typically interacted with using ENTER. For
instance, the default behavior when a link is focused and ENTER is pressed
is not to open the link in a browser but to advance or submit the form.
This commit overrides the default behavior by passing an onKeyDown prop
to the wizard which calls a custom onKeyDown function. The new custom
function does not define any behavior for when ENTER is pressed,
but does recreate DDF's default ESCAPE behavior, which is to close the
modal.
When recreating an image navigate to the review step instead of the
initial step. This requires forming the step history to enable a
functional sidebar.
Changes the Wizard's description (defined in the schema) to `Image
builder allows you to create a custom image and push it to target
environments` as indicated in the UI design prototype. The previous
description (`Create a RHEL image and push it to cloud providers`)
is no longer accurate since CentOS images are also currently supported.
Add a details step where the user can enter a descriptive image name.
This name is displayed in the review step above the Release. Both these
fields are now horizontal to save vertical space.
Update the registration options to include registering without insights.
Also, improve the text info in the step.
The org id is also removed from the registration step since we no
longer display the id in this step.
The image types 'vsphere', 'guest-image', and 'image-installer' are now
supported by the CreateImageWizard. They require no additional
customizations and support the current registration and packages
customizations.
Loading the organization ID is now done in the registration step. This
prevents the entire wizard from waiting on the org id and prevents the
wizard from remounting when the async getUser() completes.
This also mitigates the issue where clicking the image-builder nav link
would redirect you to `image-builder/image-builder` if image-builder was
already opened. And `/` didn't catch that path.
The review step is improved. The create button now states 'Create
image'. The step description text is updated. The fields to be reviewed
are now split into tabs for 'Target environment', 'Registration',
'System Configuration'. The target environments show the logo of the
environment now. The quantity of packages selected also displays now.
Tests are also updated.
The composeAdded action includes a field 'insert'. When set to true
`insert` will place the added compose at the beginning of the images
list. The test is modified so the new composes are added to the
beginning of the list.
When image creation failed to start the wizard would get stuck in the
saving state with no notice to the user. Now the user will get an error
alert contain the status code and message. The wizard no longer gets
stuck in the saving state.
Tests were failing because we are now calling getUser when wizard is shown. This call is asynchronous and requires wrapping the wizard in act to wait for it to be fulfilled
