This rule was added in rubocop-capybara 2.19.0. We were following it
about 85% of the time.
Now we won't have to check both have_css and have_selector when
searching the code.
We were getting a warning with Ruby 2.7:
```
ruby/gems/2.7.0/gems/capybara-3.37.1/lib/capybara/session.rb:377:
warning: Using the last argument as keyword parameters is deprecated;
maybe ** should be added to the call
```
On Ruby 3.0, the test failed with `Unable to find fieldset
{:text=>"Draft phase"}` and we were also getting another warning:
```
Locator Hash:{:text=>"Draft phase"} for selector :fieldset must be an
instance of String or Symbol. This will raise an error in a future
version of Capybara
```
This way we don't have to write `"spec/fixtures/files"` every time.
Note this method isn't included in factories. We could include it like
so:
```
FactoryBot::SyntaxRunner.class_eval do
include ActiveSupport::Testing::FileFixtures
self.file_fixture_path = RSpec.configuration.file_fixture_path
end
```
However, I'm not sure about the possible side effects, and since we only
use attachments in a few factories, there isn't much gain in applying
the monkey-patch.
The test was failing sometimes, probably because the "Edit" link within
the "An example legislation process" row is already present before
clicking the "All" link. This can lead to simultaneous requests.
Just removing the unnecessary click on the "All" link solves the issue.
As mentioned in the previous commits, a `<select>` field which submits
its form on change causes many accessibility and usability issues, so
we're replacing it with the order links we use everywhere else.
Since the links "Id" and "Title" by themselves don't have enough
information to let users know they're used to sort by ID or title, we
have to update them somehow. We could add a "Sort by:" prefix before the
list of links (and associate it with the `aria-labelledby` attribute);
however, we don't do this anywhere else and might look weird depending
on the screen size.
So we're simply adding "Sort by" before each link.
Now that we don't use the `wide_order_selector` partial anymore, we can
remove it alongside the styles for the `select-order` class.
When configuring phases in a process, we were validating the start date
or the end date is present, the other date is present too.
However, in other parts of the application we were checking whether a
phase is enabled and then assumed its dates were present if the phase
was enabled. However, we weren't validating this behavior, so it was
possible to enable a phase and leaving its dates blank, causing the
application to crash.
So, as suggested by Alberto, we're changing the validation rule so
phase dates are mandatory when a phase is enabled.
With this rule, the old validation rules are not necessary. I've
considered leaving them in order to avoid database inconsistencies.
However, I realized records having a disabled phase with its start and
end dates have always been valid. This means applications checking for
the presence of these dates instead of checking whether the phase is
enabled have never worked properly.
We don't have to change the logic anywhere else because as mentioned we
were already checking phases are enabled before using their dates.
The link to edit the process is already present before clicking the
"All" link, which meant the test failed sometimes because Capybara might
try to click on the "Edit" link at the same time the page is changing
due to the click on the "All" link".
Due to this issue, this test has failed at least one in our CI [1].
[1] https://github.com/consul/consul/runs/2324773853
Users don't care about database content; they care about what they see
on the screen.
Writing tests this way we also avoid potencial database inconsistencies
due to accessing the database after starting the browser.
JavaScript is used by about 98% of web users, so by testing without it
enabled, we're only testing that the application works for a very
reduced number of users.
We proceeded this way in the past because CONSUL started using Rails 4.2
and truncating the database between JavaScript tests with database
cleaner, which made these tests terribly slow.
When we upgraded to Rails 5.1 and introduced system tests, we started
using database transactions in JavaScript tests, making these tests much
faster. So now we can use JavaScript tests everywhere without critically
slowing down our test suite.
System tests are used to test the application from the user's point of
view. To test for specific exceptions, particularly regarding
authorization permissions, controller tests fit better.
Another option would be to test the page displayed shows a certain text,
like "Internal server error". I'm choosing controller tests because
they're faster and we're basically testing the same scenario many times
and we've already got a test checking what happens when users access a
page raising an exception.
Tests are easier to read now. Besides, since we changed the inputs in
the admin section so they don't use jQuery but an HTML date field,
formatting with %d/%m/%Y might not work depending on the browser's
locale.
We were repeating the same code over and over (with a few variants) to
setup tests which require an administrator. We can use a tag and
simplify the code.
In some tables, we had "actions", and some columns were also links
pointing to some places. Having both of them at the same time is
confusing, particularly since traditionally the links in the columns
pointed to the same place as some of the actions (although that's not
the case since commit 48db31cd).
We're still keeping links in tables which don't have an action column.
For instance, the proposals table has a "select" button which would be
harder to use if we had action buttons next to it.
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).
The JavaScript involved wasn't working since we removed the disable-date
attribute in commit 73ff6881.
We're also improving the JavaScript in two ways:
First, we trigger the `change` event immediately, so when the page loads
date fields are disabled when phases are disabled.
And second, we don't remove the selected dates when disabling a phase,
so disabling it and enabling it again will keep the selected values.
This is the reason why this feature was implemented in the first
place: it's easy to open the editor, make some changes, close it, and
continue without realizing the changes have not been saved.
In the rest of the forms, this functionality is quite lacking. For
starters, some forms warn if there are unsaved changes, while some forms
don't, which is highly inconsistent and disorients users.
Furthermore, we were having problems with this feature after upgrading
Turbolinks, particularly in forms using CKEditor. In these cases, a lot
of hacking needs to be done in order to make this feature work properly,
since CKEditor adds some formatting automatically, and if this is done
after the form is serialized, we'll get some unexpected behavior. On the
other hand, comparing the value of a textarea against its `defaultValue`
property will work on every edge case, including using the browser's
back button or reloading the page.
Finally, users are used to the way web forms work, and aren't used to be
asked for confirmation when they change their mind and decide to leave
the page without saving the changes. Asking them for confirmation will
be annoying in most cases. Besides that, if they accidentally leave the
page, they can use the browser's back button and they'll recover the
unsaved changes.
It's true this won't happen it they accidentally close the browser's
window, but our WatchFormChanges functionality didn't work in this case
either. Using the "beforeunload" event adds more problems than it
solves, since it doesn't support custom messages (or, to be more
precise, modern browsers ignore custom messages), and it doesn't get
along with turbolinks.
Co-Authored-By: Senén Rodero Rodríguez <senenrodero@gmail.com>
