Since forms are landmarks, screen reader users might navigate to the
form. But then they were going to find an empty form with no way to
toggle it.
Moving the button inside the form means screen reader users navigating
to the form will find the button to toggle it.
It also helps us simplifying the code; there's no need to use
data-attributes to communicate whether the form should be visible since
now we can easily use the button `aria-expanded` attribute.
We could further simplify the JavaScript if we used a CSS rule to
show/hide the form fields based on the toggle button `aria-expanded`
attribute. However, implementing the "slide" animation we use when
toggling the form with CSS is difficult and unreliable.
We were using the form and then showing it with JavaScript when advanced
search terms were present. Now we hide it with JavaScript when no
advanced search are present. This means users without JavaScript
(including users with JavaScript enabled but bad internet connections
preventing the JavaScript to load) can now access the form.
The other main difference between the two versions is the way the form
flashes while JavaScript is loading.
Previously, the form would always be hidden when no terms had been
introduced. However, when these terms were present, after submitting the
form it would briefly be hidden and then shown again.
Now the opposite happens. When advanced search terms are present, the
form is shown at all times. However, when they aren't, the form is
briefly shown before it disappears.
Here the previous behavior is arguably better because most of the time
these terms will not be present.
So basically we're significantly improving the experience of some users
at the cost of slightly worsen the experience of other users.
We're also hiding the button to show the form when JavaScript is
disabled, since in this scenario it's useless. We're using the `hidden`
attribute so hidden buttons can be detected in CSS.
Users (particularly, screen reader users) usually identify links with
things that take you somewhere, and buttons with things that either send
forms or change things on the page.
Using a button we can also use the `aria-expanded` attribute, meaning
screen reader users will know that the button has two states ("expanded"
and "collapsed"), the current state of the button, and will get
immediate feedback when clicking the button because the new state of the
button will be announced.
Thanks to this change, we can also slightly simplify the code; we
obviously have to remove the (useless) `href` attribute, and we don't
have to prevent the default event in JavaScript since there's no default
event for buttons with `type="button"`.
Note we're using both the `hidden` and `disabled` properties to
guarantee compatibility with user agents which might still display the
option even when using the `hidden` attribute or hiding it with
`display: none`. We could also use `hide()` and `show()` instead of the
`hidden` property, but since we're using the `disabled` property, I
thought the code would be easier to read if we used properties in both
cases.
Also note users will no longer be able to get, let's say, debates which
are related to goal 1 and target 2.1. We think this use case is highly
unlikely and there's no need to take it into account.
We've had to add a couple of hacks in order to make jQuery UI datepicker
work with Turbolinks, and one of our tests is failing because the
datepicker changes its height when changing from a month with 5 weeks to
a month with 6 weeks.
We could add a workaround so the test still passes (jQuery UI doesn't
provide a configuration option to always displays 6 weeks in the
datepicker), but I think it's easier to just use the HTML5 native date
input field, which also allows us to simplify the code a bit and IMHO it
improves the user experience, particularly when using mobile phones.
Since date fields are not supported in Safari and Internet Explorer,
we're still using the jQuery UI datepicker on those browsers (and on any
other browser not supporting date fields).
Due to these changes, we're moving the tests checking datepicker's
behaviour to the dashboard. I've choosing not to change the public pages
because I'm not 100% sure everybody would like this change (some people
prefer the datepicker because we can configure the way it looks).
To maintain consistency and the use of advanced search in the frontend,
we forced the expected date format for all languages. This is a temporary
solution that we should analyze in depth to allow different date formats
depending on the language in the filters.
For now we're only adding rules related to spacing and double quotes,
following the same rules we use in Ruby, which are the same rules
CoffeeScript followed when compiling these files.
We're also using the recommended ESLint rules, which will warn us about
many JavaScript common pitfalls, the `strict` rule which enforces using
strict mode, and the `no-console` rule, which will prevent us from
shipping code meant for debugging.
Although it's arguably more common to use the JSON format to define
these rules, I've chosen YAML because it's the format we use in all our
linters.
These statements were automatically added by CoffeeScript.
I'm only removing the obvious cases; there might be more cases where the
`return` statement isn't necessary.
