no issue
- adds `gh-image-uploader` that handles image uploads in a fully ember fashion and with no dependency on `uploader.js`
- adds `gh-image-uploader-with-preview` that can fully replace the old `gh-uploader`
- replace uses of `gh-uploader` in PSM & TSM with `gh-image-uploader-with-preview`
- updates the editor preview image handling to use the new `gh-image-uploader-with-preview` component
- updates the image upload modal to use `gh-image-uploader` (utilises the `saveButton=false` flag which means the preview has to be handled externally to avoid auto-replacement when typing a URL)
- removes all old `uploader.js` related code
- adds custom `RequestEntityTooLargeError` and `UnsupportedMediaTypeError` errors to our `ajax` service
no issue
- we already have similar behaviour in the `liquid-wormhole` (a dependency of `liquid-tether) but this is useful as a distinction for non-animated elements. It's also a dependency of the ember mobiledoc editor (used to render component-cards inline much the same way as this is used to render image upload components inline) and so we wouldn't really gain anything in filesize by forcing the use of `liquid-wormhole` instead
no issue
- keep the original `ember-ajax` behaviour rather than returning `false` when we hit an ajax error - we should only be using `normalizeErrorResponse` to normalize our error responses 😉
- ensures our application code has access to the returned status code for ajax errors
closes#6102
- removes the `xregexp` dependency
- pulls in the non-alphnumeric unicode list directly from the XRegExp library to ensure the word count helper still takes into account unicode strings
The XRegExp dependency was breaking as new browsers add more es6 support as evidenced in #6102. Upgrading to version 3.1.0 fixed the Chrome "experimental JS" issue but was still broken in Safari Technology Preview.
We only use one feature of `XRegExp` in a single place for a relatively non-critical feature: calculating the word count. As such I figured it may be better to drop the 63KB minified dependency and simply copy the result of the compiled regex directly until such time as XRegExp has native support in our supported browsers.
Closes#6440
* Removed the `.fake-placeholder`class from the input, test, and css
* On adding a nav item, if the url value has not been set by the user,
then set it to the base url as shown in the input
* If url has been set by the user, just do what it has always done
refs #5986
- removed static position of the skiplink to prevent pushing the page content down
- only applying :focus CSS selector to focusable sr-only content
refs #6421, #6525
- The configuration API endpoint was a bit of an animal:
- It's used currently in two ways, once for general config, another for the about page.
- These two things are different, and would require different permissions in future.
- There was also both a browse and a read version, even though only browse was used.
- The response from the browse was being artificially turned into many objects, when its really just one with multiple keys
- The new version treats each type of config as a different single object with several keys
- The new version therefore only has a 'read' request
- A basic read request with no key will return basic config that any client would need
- A read request with the about key returns the about config
- A read request with a different key could therefore return some other config
refs #6483, closes#6541
- removes unloading of all tags when entering the tags screen. Tags list will now function similarly to the content list where already-loaded tags are visible immediately.
Fixes#6483 because it prevents selectize's data being changed out from underneath it before it's destroyed.
closes#6512
- because of the keyForAttribute method in the application serializer, the hash property must be snake_case
- change the normalize method to the more concise normalizeHash method available on the rest serializer
