Just like we add the `storage_` prefix for new records so we can use
both Active Storage and Paperclip at the same time.
Now the migration actually works, at least for basic cases.
`dalli_store` is deprecated since dalli 2.7.11.
We can now enable cache_versioning. We didn't enable it when upgrading
to Rails 5.2 because of possible incompatibility with `dalli_store` [1],
even though apparently some the issues were fixed in dalli 2.7.9 and
dalli 2.7.10 [2].
Since using cache versioning makes cache expiration more efficient, and
I'm not sure whether the options we were passing to the dalli store are
valid with memcache store (documentation here is a bit lacking), I'm
just removing the option we used to double the default cache size on
production.
[1] https://www.schneems.com/2018/10/17/cache-invalidation-complexity-rails-52-and-dalli-cache-store
[2] https://github.com/petergoldstein/dalli/blob/master/History.md
Since version 2.0 introduced many breaking changes, we're upgrading to
it first.
The changes have been done by installing the rubocop-faker gem and
running:
```
rubocop \
--require rubocop-faker \
--only Faker/DeprecatedArguments \
--auto-correct
```
Our previous system to delete cached attachments didn't work for
documents because the `custom_hash_data` is different for files created
from a file and files created from cached attachments.
When creating a document attachment, the name of the file is taken into
account to calculate the hash. Let's say the original file name is
"logo.pdf", and the generated hash is "123456". The cached attachment
will be "123456.pdf", so the generated hash using the cached attachment
will be something different, like "28af3". So the file that will be
removed will be "28af3.pdf", and not "123456.pdf", which will still be
present.
Furthermore, there are times where users choose a file and then they
close the browser or go to a different page. In those cases, we weren't
deleting the cached attachments either.
So we're adding a rake task to delete these files once a day. This way
we can simplify the logic we were using to destroy cached attachments.
Note there's related a bug in documents: when editing a record (for
example, a proposal), if the title of the document changes, its hash
changes, and so it will be impossible to generate a link to that
document. Changing the way this hash is generated is not an option
because it would break links to existing files. We'll try to fix it when
moving to Active Storage.
When we assigned `I18n.available_locales = default_locales` in the
`ensure` block, `I18n.locale` was set to `:zh-TW`, which is not one of
the default locales.
In some cases this resulted in tests failing:
```
I18n::InvalidLocale:
:"zh-TW" is not a valid locale
```
Note we're making the validation rule dynamic so it's affected by the
way we stub the constant in the tests to emulate data created in old
applications.
Co-Authored-By: Javi Martín <javim@elretirao.net>
So tests won't fail when an institution changes the default organization
name.
The tests are also easier to understand now, since it's more obvious
where the "CONSUL" text is coming from.
By using real XML responses developers will be able to understand better
how the integration works (the data flow), and the correspondency between
`remote_census` settings and their place at a real XML response.
As `stubbed_responses` methods were removed from the model layer now the
stubbing part should be managed from the test environment code so also
added a new helper module `RemoteCensusSetup` that can be used anywhere
where we need to call the web service.
Co-Authored-By: Javi Martín <javim@elretirao.net>
By simplyfing the responses the configuration for specs can be simpler too.
We're also using more generic terms instead of the ones used in Madrid's
Census API.
Co-Authored-By: Javi Martín <javim@elretirao.net>
Otherwise the variants returned for document_type="1" and
document_number="" will be
`["0", "00", "000", "0000", "00000", "000000", "0000000", "00000000"]`
which seems to be useless.
Probably this case is not real for production environments where those
arguments will always be fullfilled but seems to be interesting for
testing environment where this method is being called when those
paremeters where empty.
This rule was added in Rubocop 0.91.0. A similar rule named
LeakyConstantDeclaration was added in rubocop-rspec 1.34.0.
Note using the FILENAMES constant did not result in an offense using the
ConstantDefinitionInBlock rule but did result in an offense using the
LeakyConstantDeclaration rule. I've simplified the code to get rid of
the constant; not sure why we were adding a constant with `||=` in the
middle of a spec.
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()
The method `available_locales` in
`RemoteTranslations::Microsoft::AvailableLocales` needs to execute a
request to an external server in order to work, meaning it will fail if
the machine its executed on doesn't have an internet connection.
So we're stubbing the method in the tests using it.
Implementation tries to be open for further extensions, such as deciding on
search dictionary based on configuration option or by locale set for
given user.
If we didn't run this task, investments for existing budgets wouldn't
show their administrator/valuators as an option when we're editing them,
leading to data loss.
We were manually doing the same thing, generating inconsistent results,
since the method `valuation_tag_list` was using the `valuation` context,
when actually the expected behavior would be to use the `valuation_tag`
context.
This method is ambiguous. Sometimes we use it to set invalid data in
tests (which can usually be done with `update_column`), and other times
we use it instead of `update!`.
I'm removing it because, even if sometimes it could make sense to use
it, it's too similar to `update_attributes` (which is an alias for
`update` and runs validations), making it confusing.
However, there's one case where we're still using it: in the
ActsAsParanoidAliases module, we need to invoke the callbacks, which
`update_column` skips, but tests related to translations fail if we use
`update!`. The reason for this is the tests check what happens if we
restore a record without restoring its translations. But that will make
the record invalid, since there's a validation rule checking it has at
least one translation.
I'm not blacklisting any other method which skips validations because we
know they skip validations and use them anyway (hopefully with care).
We were very inconsistent regarding these rules.
Personally I prefer no empty lines around blocks, clases, etc... as
recommended by the Ruby style guide [1], and they're the default values
in rubocop, so those are the settings I'm applying.
The exception is the `private` access modifier, since we were leaving
empty lines around it most of the time. That's the default rubocop rule
as well. Personally I don't have a strong preference about this one.
[1] https://rubystyle.guide/#empty-lines-around-bodies
Having exceptions is better than having silent bugs.
There are a few methods I've kept the same way they were.
The `RelatedContentScore#score_with_opposite` method is a bit peculiar:
it creates scores for both itself and the opposite related content,
which means the opposite related content will try to create the same
scores as well.
We've already got a test to check `Budget::Ballot#add_investment` when
creating a line fails ("Edge case voting a non-elegible investment").
Finally, the method `User#send_oauth_confirmation_instructions` doesn't
update the record when the email address isn't already present, leading
to the test "Try to register with the email of an already existing user,
when an unconfirmed email was provided by oauth" fo fail if we raise an
exception for an invalid user. That's because updating a user's email
doesn't update the database automatically, but instead a confirmation
email is sent.
There are also a few false positives for classes which don't have bang
methods (like the GraphQL classes) or destroying attachments.
For these reasons, I'm adding the rule with a "Refactor" severity,
meaning it's a rule we can break if necessary.
