We were using `Setting["url"]` to verify the content belonged to the
application URL, but we can use `root_url` instead.
Note that means we need to include the port when filling in forms in the
tests, since in tests URL helpers like `polymorphic_url` don't include
the port, but a port is automatically added when actually making the
request.
This way the code is easier to follow; the code checking whether the
list has contents is in the partial rendering the list.
We also remove some duplication setting up related content in the
controllers.
For some reason, we have to manually ignore i18n keys which were
automatically ignored when the code was in the view.
"Is it related content? Yes (link) No (link)".
After reading that, you probably haven't decided whether to click the
"Yes" link or the "No" link or neither of them. You probably have to
read the content it's related to, and then go back to these links.
That's the reason we style them so they're on the right of the screen on
languages which are read from left to right, so first we read the
content and then we read the links to mark it as related or unrelated.
So it makes sense to put these links after the content they refer to.
This way users with small screens as well as screen reader users will
also see/hear these links after they get the proper context. It will
also be more intuitive for keyboard users, who saw the focus on the
links before focusing on the title of the related content, when they
probably expected the opposite.
Since we're now using a flexbox layout instead of a right float, this
also means the content will automatically be shown on the left when
switching the content direction to "right to left".
We were only showing these actions to users with small screens and to
mouse users on hover. Keyboard users or users with a touch screen of a
medium or large size could never find out the actions were there.
It didn't have a `for` attribute and so it wasn't correctly associated
with its input. That means clicking on / touching the label didn't have
the effect of focusing on the field, and screen readers wouldn't
announce the label.
The button was announced as expanded when the form was hidden and as
collapsed when the form was shown.
This is because Foundation sets the expanded attribute based on whether
the class to toggle already exists. Since initially the form had the
"hide" class and the button toggled that class, Foundation checked that
the class was already present and so set the button as expanded.
So we're changing the toggler class for a class we don't use at all,
just so Foundation initiall sets `aria-expanded=false` and then changes
it to `aria-expanded=true` after the button is clicked. Then we're
ignoring this class completely and are styling this form with CSS
instead.
We could also use a toggler class like "visible" and write something
like:
```
.add-related-content + form:not(.visible) {
display: none;
}
```
However, using ARIA attributes is more robust as it guarantees the
styles will always be in sync with what screen reader users experience.
And we could also remove all the Foundation toggler functionality and
use our own JavaScript to handle the button state. We might do so in the
future.
This way the relationship between the two elements is more obvious. And
since now the button and the form are siblings, it's easier to find one
based on the other using CSS or JavaScript.
A button cannot be inside an anchor tag, and it might confuse some
browsers or screen readers.
We're also making it clear in the tests that the intention is to use a
button there by using `click_button` instead of `click_on` since the
latter also clicks on links.
