We didn't upgrade Turbolinks when we upgraded to Rails 5 so we didn't
upgrade too many things at the same time, and postponed it... until now
:).
Note upgrading Turbolinks fixes an issue with foundation's sticky when
using the browser's back and forward buttons. We're adding tests for
these scenarios.
Co-authored-by: Senén Rodero Rodríguez <senenrodero@gmail.com>
In the past we had huge problems trying to make it work with Turbolinks.
However, after updating foundation-rails in commit 58071fd6, these hacks
aren't necessary anymore.
We're adding a test for the scenario of visiting a page using
Turbolinks, which was missing, so we're sure we aren't breaking
anything.
Note the sticky will still not work after using the browser back button.
We haven't been able to make it work with turbolinks-classic; we'll fix
this issue when upgrading turbolinks.
Its known that initializing a map when it is inside a hidden element
wont work when hidden element is shown, so its makes sense to
avoid initialization of hidden maps.
When a map lives within a hidden layer we need to initialize the
map after the event of showing that hidden layer, in our case when
admin settings tab is shown.
When a legislation process is deleted, everything related will be
deleted, including the answers. This `dependent: :destroy` was causing
that users accounts were being accidentally deleted.
The test wasn't working when postgres used the English dictionary
because in English the word "what" was ignored (or, at least, not given
enough relevance) while searching. When we wrote the test, it passed
because back then we always used the Spanish dictionary. However, when
we switched to a dictionary based on the default locale (in commit
d99875cd), we had to force this test to keep using the Spanish
dictionary.
Using the Spanish dictionary in a test where all texts are in English is
strange to say the least ;). So here we're making the test a bit easier
to understand.
Since now we're only using the `:spanish_search` tag in one test, I've
decided to remove it and simply add it to that test's setup.
We were getting a deprecation message in Rails 5.2:
The missing? predicate is deprecated and will be removed in Rails 6.0.
Please use not_found? as provided by Rack::Response::Helpers
With two concurrent requests, it's possible to create two ballot lines
when only one of them should be created.
The reason is the code validating the line is not thread safe:
```
if ballot.amount_available(investment.heading) < investment.price.to_i
errors.add(:money, "insufficient funds")
end
```
If the second request executes this code after the first request has
executed it but before the first request has saved the record to the
database, both records will pass this validation and both will be saved
to the database.
So we need to introduce a lock. Now when the second request tries to
lock the ballot, it finds it's already locked by the first request, and
will wait for the transaction of the first request to finish before
checking whether there are sufficient funds.
Note we need to disable transactions during the test; otherwise the
second thread will wait for the first one to finish.
Also note that we need to update a couple of tests because records are
reloaded when they're locked.
In one case, reloading the ballot causes `ballot.user` to be `nil`,
since the user is hidden. So we hide the user after creating all its
associated records (which is the scenario that would take place in real
life).
In the other case, reloading the ballot causes `ballot.user` to reload
as well. So we need to reload the user object used in the test too so it
gets the updates done on `ballot.user`.
I haven't been able to reproduce this behavior in a system test. The
following test works with Rails 5.0, but it stopped working when we
moved to system tests in commit 9427f014. After that commit, for reasons
I haven't been able to debug (reintroducing truncation with
DatabaseClaner didn't seem to affect this test, and neither did
increasing the number of threads in Puma), the two AJAX requests
executed here are no longer simultaneous; the second request waits for
the first one to finish.
scenario "Race conditions with simultaneous requests", :js do
allow_any_instance_of(Budget::Ballot::Line).to receive(:check_sufficient_funds) do |object|
allow(object).to receive(:check_sufficient_funds).and_call_original
object.check_sufficient_funds
sleep 0.3
end
["First", "Second"].each do |title|
create(:budget_investment, :selected,
heading: california,
price: california.price,
title: title
)
end
login_as(user)
visit budget_investments_path(budget, heading_id: california.id)
within(".budget-investment", text: "First") { click_link "Vote" }
within(".budget-investment", text: "Second") { click_link "Vote" }
expect(page).to have_link "Remove vote"
expect(Budget::Ballot::Line.count).to eq 1
end
Rails 5.2 crashes in the `db:create` task because it tries to run the
`after_initialize` block before the database is created.
The easiest way to solve it is to move the code out of the initializer
and calculate the API type definitions on demand. Note results are still
cached using a class instance variable (not to be confused with a class
variable), and so once definitions are obtained, they will remain
constant until the application is restarted, even in the development
environment.
These tasks are not needed for new installations, and in existing
installations they've already been executed when upgrading to version
1.1.
One of them also raises a warning in Rails 5.2:
DEPRECATION WARNING: Dangerous query method (method whose arguments are
used as raw SQL) called with non-attribute argument(s): "MIN(id) as id".
Non-attribute arguments will be disallowed in Rails 6.0. This method
should not be called with user-provided values, such as request
parameters or model attributes. Known-safe values can be passed by
wrapping them in Arel.sql()
We were treating legislation proposals as if they were proposals,
omitting the "legislation" namespace, and so we were flagging/unflagging
proposals when we wanted to flag/unflag a legislation proposal.
We weren't using `foundation()` in these cases, so after flagging a
debate or a comment, we had to reload the page before we could unflag
it.
We're also adding a test for the fix in commit ea85059d. This test shows
it's necessary to filter the elements with JavaSctipt using `first()` if
we want the same code to work with comments.
Co-Authored-By: taitus <sebastia.roig@gmail.com>
We don't need to set this value. In commit f2ef27d3 I made a mistake
thinking `Globalize.locale` and `I18n.locale` should always be in sync,
but they're actually automatically in sync when `Globalize.locale` is
`nil`.
So the best way to avoid any issues is not to assign `Globalize.locale`,
and use `Globalize.with_locale` where necessary instead.
These routes are solved in a different way because of an inconsistency:
we define `groups` and `budget_investments`; we should either use the
`budget_` prefix in all places or remove it everywhere.
We can now share code using `polymorphic_path` even with these models.
In the past, we couldn't use `polymorphic_path` in many places. For
instance, `polymorphic_path(budget, investment)` would return
`budget_budget_investment_path`, while in our routes we had defined
`budget_investment_path`.
With the `resolve` method, introduced in Rails 5.1, we can use symbols
to define we want it to use `investment` instead of `budget_investment`.
It also works with nested resources, so now we can write
`polymorphic_path(investment)`.
This makes the code for `resource_hierarchy_for` almost impossible to
understand. I reached this result after having a look at the internals
of the `resolve` method in order to get its results and then remove the
symbols we include.
Note using this method will not make admin routes compatible with
`polymorphic_path`. Quoting from the Rails documentation:
> This custom behavior only applies to simple polymorphic URLs where a
> single model instance is passed and not more complicated forms, e.g:
> [example showing admin routes won't work]
Also note that now the `admin_polymorphic_path` method will not work for
every model due to inconsistencies in our admin routes. For instance, we
define `groups` and `budget_investments`; we should either use the
`budget_` prefix in all places or remove it everywhere. Right now the
code only works for items with the prefix; it isn't a big deal because
we never call it with an item without the prefix.
Finally, for unknown reasons some routing tests fail if we use
`polymorphic_path`, so we need to redefine that method in those tests
and force the `only_path: true` option.
With chromedriver >= 80, the tests are freezing sometimes, particularly
when the same editor is loaded again.
We don't know whether it's a CKEditor issue or a chromedriver issue. In
the past we've had some errors related to CKEditor trying to load the
same instance twice and we aren't sure they have been fixed since we
could never reproduce them.
It could be a coincidence, though. If we modify the views so the only
content of the `<body>` tag is a textarea with the `html-area` class,
chromedriver freezes even if we only access the page once. So maybe
we're only detecting the problem on the second visit because the second
request is faster than the first one.
Since chromedriver no longer hangs after this change, we don't have to
force any chromedriver version anymore.
After upgrading to chromedriver 80, tests checking CKEditor's content
were causing chromedriver to hang. That's why we were configuring
webdrivers to use an older chromedriver.
Version 80 of chromedriver introduced several issues regarding frames.
Debugging shows in this case chromedriver froze when we used `setData`
and then `within_frame`. Since adding a `sleep` call made it work, we
think `within_frame` was being executed before `setData` had finished.
The fact that `setData` causes the browser to enter the frame having
CKEditor is probably the reason.
Even though the `setData` method provides a callback when it's finished,
configuring it so the rest of the Ruby code isn't executed until that
happens leads to complex code. Using Capybara's `set` to fill in the
editor is IMHO a bit easier to understand.
After this change, since we're using a method provided by Capybara
instead of executing asynchronous JavaScript code, we don't have to
check CKEditor has been filled anymore. The "Admin Active polls add"
test, which failed on my machine without that check, now passes.
If we don't use the `exact` option, tests will pass even if filling in
CKEditor adds the content twice or adds the new content to the existing
content, which has actually happened and has gone mostly unnoticed while
testing several ways to fill in CKEditor with Capybara (particularly,
when using Capybara's `send_keys` method). The problem was detected by
just one test, which checked the original content wasn't present anymore
after updating a record.
When a user replies to a comment whose responses was hidden at the
moment of reply form submission and although the reply were correctly
added to the DOM it was hidden because was added to a collapsed list.
This solution is about showing all responses of parent comment after adding
the new comment to the DOM so the user can see new reply into the screen.
(This is not applicable to root comments which cannot be collapsed)
Extract the needed portion of code to a new partial to be able to update
only the elements needed when a new comment is added keeping UI properly
updated.
In this case the confirmation dialog isn't really necessary since the
action to enable/disable the setting can easily be undone.
Furthermore, these tests were failing with Chrome 83, probably because
we use `confirm_dialog` and then we use `visit` without checking the
page in between.
In theory we shouldn't need to check the page in between because the
request generated by `confirm_dialog` is a synchronous one and so
`visit` isn't executed after the previous request has finished, but
apparently this behavior has changed in Chrome 83.
We could add an expectation before executing the `visit` method, but
that wouldn't improve the usability of the application.
The latest stable version is causing problems on some machines, hanging
forever in tests involving frames. So we're installing an old version
which works with the latest Chrome.
Note this means we're using an unsupported version. Officially, only the
latest chromedriver supports the latest Chrome.
We're using 2.38 instead of a more recent one (like 2.40) because it's
the one we specified in our Dockerfile.
See also:
https://bugs.chromium.org/p/chromedriver/issues/detail?id=3361
The number of errors in a form includes several errors for the same
field. For example, if a title is mandatory and has to have at least 5
characters, leaving the title blank will result in two errors. So users
will be invited to look for two errors, but they'll only find one field
with errors.
So it's a bit more intuitive to show as many errors as fields having
errors.
Note we're excluding errors on `:base`, which is a bit of a hack for
errors in association fields. For example, if the title of one
translation is not present, `resource.errors.messages` will contain two
elements: one for the translation's title, and one for the `base` field.
This resulted in the count of errors being 2 when there was only one.
Also note I haven't found a way to count errors on all `has_many`
relations. That is, if two translations have a missing title field, only
one error will be mentioned in the message (as it did before this
commit).
We were submitting the form without checking the AJAX request to attach
the image had finished, so sometimes two requests were executed at the
same time. Sometimes this made InvisibleCaptcha to go crazy and report
the form was submitted too quickly.
Checking the first AJAX request has finished before submitting the form
solves the problem.
We were checking `expect_document_has_title(0, "My Title")`, which was
already true before the AJAX request generated by `attach_file` had
finished.
That meant the AJAX request sometimes was handled after this test had
finished, affecting the following test and causing it to fail because
its cookie was overwritten and so `current_user` was set to `nil`.
In the test checking the filename is present, a similar scenario was
taking place: we were updating the `.file-name` element in the `change`
event of `fileupload` (using `App.Documentable.setFilename`); that is,
when the AJAX request started. And so the test passed before the request
was finished, causing the same issue.
