Since we were defining the selection to have the same text color and
background color as the element they were in, it resulted in the
selection being invisible.
It wasn't that noticeable because we were using this color combination
mainly in links and buttons, and selecting their text is not as common.
But we plan to use the `$brand` color on budget headers as well, and
this issue is more obvious there.
Browsers like Chrome weren't that affected because they automatically
make the selection semi-transparent and so the selected text still had a
slightly different color. In order to prevent this effect when the
selection is white, we're using a 0.99 opacity (in this case Chrome
ignores numbers higher that 0.998).
Setting the color to `$white` or `#fff` while setting the background to
`$brand` is a pattern we were using in many places. Since we're going to
edit it in order to fix the `::selection` behavior, we're defining the
pattern in one place.
One file was OK when we only had a couple of mixins, but recently we've
been adding quite a few.
We can now avoid a SCSS Lint warning by excluding just the file with the
affected mixin.
Although our installer automatically generates a secret key base, we
want to avoid any chance of installations who don't use the installer
having an empty secret key base or using the default secret key base we
provide to use on development and test environments.
Due to a bug in dependabot (see issue 2198 in the
`dependabot/dependabot-core` repository), Dependabot would open pull
requests on every fork, which would be inconvenient.
So we're disabling the updates until this issue is fixed; since we
currently have almost 50 open pull requests, it shouldn't affect us that
much.
Since the `@ballot_referer` variable was only set in the lines
controller, it didn't work when we accessed the ballot page without
adding a line.
Note it still doesn't work if we access the ballot page directly by
entering the URL in the browser's address bar.
Even if we usually only access these pages for the current budget, that
might not always be the case, and now that we've unified budget landing
pages, there's no point in them pointing to the index anymore.
There was a big difference between the current budget and a specific
budget landing page. This didn't really make too much sense. Also, it
was not possible to know how a draft participatory budget will look
before it was published.
By unifying those two views now they will look quite similar and it
will be possible for administrators to preview any draft budget and to
know how the budget will look like before actually publishing it.
It was added because a test failed without turbolinks. However, writing
the test so it doesn't update the database at the same time the browser
is doing a request also solves the problem and makes the test more
robust.
Using the name instead of using the name and the price is IMHO more
consistent with the rest of the application, particularly for screen
reader users. Writing texts clicking those links is also easier.
I think the main reason why we used the price as part of the link was so
the clickable area was bigger. We can accomplish the same result with
CSS.
Using flex we don't have to rely on JavaScript to equalize the item.
Besides, we've had problems with JavaScript in the past.
We're also adjusting the width of the elements; previously, even though
we defined a width of 16.666% for each element, only five elements would
be on the same row. It happenend because these elements were styled with
inline-block and the generated HTML contained a newline character
between <li> tags, meaning a space character was introduced between
elements. The width of the mentioned space character wasn't being taken
into account when calculating the width.
Using flex, there's no space character between items and we have to
define the margin between them. We're taking this margin into account
when calculating the width.
Now it's easier to change the investments filter. Previously we had to
go back to the budget index page, change the filter there, and then
select one heading.
Now the links to change the current filter in the budget index page
aren't needed anymore.
We weren't using the "Feasible" filter anywhere in the
application, and it's hard for users to know the difference between
"Feasible" and "Not unfeasible".
Apart from 'balloting', there are more phases between
'publishing_prices' and 'finished'. So, it doesn't make sense that the
filter 'selected' is only applied to those two. With this change the
filter will be applied as follow:
'not_unfeasible' for phases:
- informing
- accepting
- reviewing
- selecting
- valuating
selected for phases:
- publishing_prices
- balloting
- reviewing_ballots
winners for phases:
- finished
We were defining the same filters in three different controllers. We
were also adding a method in the ApplicationController which only made
sense in the same three controllers.
Since we were using an icon font with no text, screen readers were
announcing things like "Enabled, L", trying to read the icon generated
with CSS.
Using text and replacing it with CSS with an icon solves the problem.
We could also use aria-label, but I prefer using "Yes/No" so the text
can be shown/hidden with CSS. Also useful when using
`save_and_open_page` during tests, since the displayed page will not
have any CSS rules applied.
Out of several existing techniques to hide text [1], we're setting the
font size to 1px in combination with moving the content off-screen
because that way we can override it in the `::before` element.
Just moving the content off-screen has the inconvenient of the content
still being taken into account when calculating the text indentation.
And just using a 1px font would make a 1px-sized square appear when
selecting text, which could confuse users.
[1] https://webaim.org/techniques/css/invisiblecontent/
Using 10000px means content will be visible on screens with ultra-high
resolution (although I don't think they exist nor will exist anytime
soon).
Having screens in the future with a width of 1000rem is less likely,
since a line with 1000 characters would be pretty much impossible to
read.
