New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Fix checking case sensitivity when value is case insensitive #789

Closed
wants to merge 3 commits into
base: master
from

Conversation

Projects
None yet
4 participants
@albertyw
Contributor

albertyw commented Oct 4, 2015

I ran into the same issue as #787 and fixed it here.

context 'when the value has no case sensitivity' do
it 'accepts', focus: true do
model = define_model_validating_uniqueness(
attribute_type: :string

This comment has been minimized.

@houndci-bot

houndci-bot Oct 4, 2015

Put a comma after the last parameter of a multiline method call.

@@ -408,6 +408,9 @@ def validate_case_sensitivity?
if value.respond_to?(:swapcase)
swapcased_value = value.swapcase
if swapcased_value == value
swaped_value += '_case'

This comment has been minimized.

@houndci-bot

houndci-bot Oct 4, 2015

Useless assignment to variable - swaped_value.

@@ -408,6 +408,10 @@ def validate_case_sensitivity?
if value.respond_to?(:swapcase)
swapcased_value = value.swapcase
if swapcased_value == value
swapcased_value += '_CASE'
value += '_case'

This comment has been minimized.

@houndci-bot

houndci-bot Oct 4, 2015

Useless assignment to variable - value.

@@ -408,6 +408,10 @@ def validate_case_sensitivity?
if value.respond_to?(:swapcase)
swapcased_value = value.swapcase
if swapcased_value == value
existing_record.update(@attribute => value+'_case')

This comment has been minimized.

@houndci-bot

houndci-bot Oct 4, 2015

Surrounding space missing for operator '+'.

@tovodeverett

This comment has been minimized.

tovodeverett commented Oct 4, 2015

What if there are constraints on the value that make appending "_case" illegal?

@@ -450,9 +450,19 @@
expect(record).to validate_uniqueness
end
context 'when the value has no case sensitivity' do

This comment has been minimized.

@mcmire

mcmire Oct 4, 2015

Collaborator

Maybe "when the value is the same as its swapcased version"?

Also add a line break above this line, per style.

attribute_type: :string,
validation_options: { case_sensitive: true },
)
record = create_record_from(model, attribute_name => '123')

This comment has been minimized.

@mcmire

mcmire Oct 4, 2015

Collaborator

I think attribute_name should be value, right?

This comment has been minimized.

@albertyw

albertyw Oct 5, 2015

Contributor

I thought that too, but changing that line to record = create_record_from(model, value: '123') makes the test fail.

This comment has been minimized.

@mcmire

mcmire Oct 5, 2015

Collaborator

Oh... gah... I was skimming over the code and wasn't looking closely. The key is the name of an attribute, so this is right.

end
context 'when case_insensitive is specified' do
context 'when case_sensitive is specified' do

This comment has been minimized.

@mcmire

mcmire Oct 4, 2015

Collaborator

This line shouldn't be changed, "specified" is referring to the matcher. (Really this should say "when qualified with case_insensitive", but I'll fix it later.)

@mcmire

This comment has been minimized.

Collaborator

mcmire commented Oct 4, 2015

The current behavior is to test that the record either allows or disallows the swapcased version of the value whether case_insensitive is specified or not. I just read that issue and I think what we should do is not even bother making this assertion if the value is the same as its swapcased version instead of adding alpha characters onto the value (or raising an error as the aforementioned issue suggests), because in this case, the case_sensitive: false setting on the validation really isn't doing anything.

So what if you changed the validate_case_sensitivity? method to look like this?

if value.respond_to?(:swapcase) && value != value.swapcase
  # allow or disallow
else
  true
end
@tovodeverett

This comment has been minimized.

tovodeverett commented Oct 4, 2015

From a testing standpoint, the default behavior is to verify that the uniqueness is indeed case sensitive (or to test that the uniqueness is indeed case insensitive if we declare case_insensitive). If I make a mistake and supply a model that has a value that is swapcase immutable, then we're no longer testing either of those, but as a user I wouldn't necessarily be aware of that change since it was made silently.

Perhaps we need an option for ignore_case_sensitivity. For instance, if I have a string where I store formatted phone numbers. I might have validation rules that make it impossible to store alpha characters. On the other hand, I might have a field where I want to test that it is case insensitive, but I was oblivious and set the value to '123'. It would be nice in the latter case to be informed, "Hey, if you want the test to validate case sensitivity, you had better pass a value that lets us test that!"

@mcmire

This comment has been minimized.

Collaborator

mcmire commented Oct 4, 2015

@tovodeverett Okay, that's fair. I don't know if we need an option for it because then you'd have to remember to use that too. But that's a good warning to have.

I'm trying to understand your original use case... in your test, did you pre-create a record before using the matcher?

@albertyw

This comment has been minimized.

Contributor

albertyw commented Oct 4, 2015

In my tests, I was not pre-creating the object (shoulda-matchers does it internally though).

I think @tovodeverett's suggestion is best. The current status will fail with a non-useful error and there's no way modifying the value will work because of possible additional validations. Instead of defaulting to true if the value isn't case sensitive (which therefore doesn't test anything), perhaps a RuntimeError can be raised so that an extra option doesn't need to be added.

@mcmire

This comment has been minimized.

Collaborator

mcmire commented Oct 4, 2015

Okay, so you're saying that if you as the developer accidentally use a value that wouldn't create a good test by being un-swapcase-able, then you'd want the test to fail so that you're forced to replace it with something better?

@albertyw

This comment has been minimized.

Contributor

albertyw commented Oct 4, 2015

Well, I guess to expand a bit, if there are validations to preclude changing the value to something swapcase-able (e.g. a phone number), then the test should be changed to be case insensitive. (i.e. if you hit 1, then an error message should appear to change your code to 2 or 3)

  1. Validation - sensitive, test - sensitive, value - insensitive = throw error
  2. Validation - sensitive, test - insensitive, value - insensitive = accepts
  3. Validation - sensitive, test - sensitive, value - sensitive = accepts (works already)
@tovodeverett

This comment has been minimized.

tovodeverett commented Oct 4, 2015

In my case, the underlying value is case-sensitively unique. It's a field where I store pattern matches (SQL patterns for LIKE). I'm using factory_girl, and the pattern was along the lines of '%-1-%', which is, of course, unchanging with swapcase! I realized that since my underlying implementation was case-sensitively unique, I didn't want to indicate to the test that it should be case insensitive (since that didn't match the implementation). So I changed the value in factory_girl to '%-1[-a]%' and thus ensured that I got a test that it was indeed case sensitive.

I'm 95% onboard with albertyw's suggestion - if the stored value cannot contain alpha characters, then suggest to the developer that they add case_insensitive to the test. Basically, the error message would guide the developer into one of two resolutions: Either supply a test value that mutates with swapcase, or explicitly make the test case_insensitive.

@mcmire

This comment has been minimized.

Collaborator

mcmire commented Oct 4, 2015

Okay, just so I understand, it seems like there are two uses cases here:

  1. The developer has a case-sensitive validation in the model and is using the matcher without case_insensitive, but the value the matcher is using isn't swapcase-able. To fix this, the developer should provide a record with a value that is swapcase-able.
  2. The developer has a case-sensitive validation on an attribute that holds values which do not contain any alpha characters. The value the matcher is using also isn't swapcase-able, but changing the value won't work -- in this case the developer needs to change the validation (and the matcher) to be case-insensitive.

So as you say @tovodeverett we just need to explain to the developer why is this happening and guide them into one of these choices.

@tovodeverett

This comment has been minimized.

tovodeverett commented Oct 4, 2015

Yup! Technically case 2 is a simplification, but I think it's good enough.

Technically, the attribute in question isn't really case-sensitive or case-insensitive - the concept of case sensitivity doesn't apply to an attribute which can't store alpha characters, but since calling it case-insensitive will work, I think that's a good solution (and definitely simpler than defining yet another concept). Don't get me started on I18N, Turkish i's, and collations!

validation_options: { case_sensitive: true },
)
record = create_record_from(model, attribute_name => '123')
expect{validate_uniqueness.matches?(record)}.to raise_error Shoulda::Matchers::ActiveModel::CouldNotValidateCaseError

This comment has been minimized.

@houndci-bot

houndci-bot Oct 5, 2015

Line is too long. [127/80]
Space missing to the left of {.
Space missing inside {.
Space missing inside }.

)
record = create_record_from(model, attribute_name => '123')
case_error = Shoulda::Matchers::ActiveModel::CouldNotValidateCaseError
expect { validate_uniqueness.matches?(record) }.to raise_error case_error

This comment has been minimized.

@houndci-bot

houndci-bot Oct 5, 2015

Line is too long. [83/80]

)
record = create_record_from(model, attribute_name => '123')
case_error = Shoulda::Matchers::ActiveModel::CouldNotValidateCaseError
expect {

This comment has been minimized.

@houndci-bot

houndci-bot Oct 6, 2015

Avoid using {...} for multi-line blocks.

@albertyw

This comment has been minimized.

Contributor

albertyw commented Oct 6, 2015

@tovodeverett @mcmire I've updated the pull request based on your comments.

@mcmire

This comment has been minimized.

Collaborator

mcmire commented Oct 6, 2015

Cool, thanks so much. I'll take a look at this tonight.

@mcmire

This comment has been minimized.

Collaborator

mcmire commented Oct 6, 2015

@albertyw @tovodeverett Okay, I reworded the error message to explain this in more detail. What do you think about this? (the model name, attribute, and value will be filled in dynamically)

Your Example model has a uniqueness validation on :attr which is declared to be case-sensitive, but the value the uniqueness matcher used, "123", doesn't contain any alpha characters, so using it to test the case-sensitivity part of the validation is ineffective. There are two possible solutions for this depending on what you're trying to do here:

  1. If you meant for the validation to be case-sensitive, then you need to give the uniqueness matcher a saved instance of Example with a value for :attr that contains alpha characters:

    # RSpec
    it do
      record = Example.create!(attr: "some value with alpha characters")
      expect(record).to validate_uniqueness_of(:attr)
    end
    
    # Minitest (Shoulda)
    should "validate uniqueness of :attr" do
      record = Example.create!(attr: "some value with alpha characters")
      assert_accepts validate_uniqueness_of(:attr), record
    end
  2. If you meant for the validation to be case-insensitive, then you need to qualify the validation, as well as the matcher, appropriately:

    # Model
    class Example < ActiveRecord::Base
      validates_uniqueness_of :attr, case_sensitive: false
    end
    
    # RSpec
    describe Example do
      it { should validate_uniqueness_of(:attr).case_insensitive }
    end
    
    # Minitest (Shoulda)
    class ExampleTest < ActiveSupport::TestCase
      should validate_uniqueness_of(:attr).case_insensitive
    end
@tovodeverett

This comment has been minimized.

tovodeverett commented Oct 7, 2015

I think that error message is insanely awesome!!!! In fact, it's so awesome that I'm going to purposely use a few test values that don't have alpha characters just so I can smile when it prints out! 👍

It does a great job of explaining to the developer what the problem is, what the alternatives are, and provides a good example of how to resolve the situation under each alternative.

Be careful - if every error message I got as a developer was this helpful, Stack Overflow would need a new business model!

@albertyw

This comment has been minimized.

Contributor

albertyw commented Oct 7, 2015

I like the error message too - it's very clear. However, I think it's a bit long to output to console (certainly can work, but at minimum the formatting would be stripped). Should the error message link to the text in the readme or the website?

@mcmire

This comment has been minimized.

Collaborator

mcmire commented Oct 7, 2015

@tovodeverett Thanks for the kind words! 😊

@albertyw Yeah... I was starting to think that. I do love great error messages, but if you saw four of these in a row then it might be a bit much. Let's try this:

Your Example model has a uniqueness validation on :attr which is declared to be case-sensitive, but the value the uniqueness matcher used, "123", doesn't contain any alpha characters, so using it to test the case-sensitivity part of the validation is ineffective. There are two possible solutions for this depending on what you're trying to do here:

  1. If you meant for the validation to be case-sensitive, then you need to give the uniqueness matcher a saved instance of Example with a value for :attr that contains alpha characters.
  2. If you meant for the validation to be case-insensitive, then you need to add case_sensitive: false to the validation and add case_insensitive to the matcher.

For full examples, please see:

http://github.com/thoughtbot/shoulda-matchers/wiki/error-uniqcs

@@ -412,6 +412,9 @@ def validate_case_sensitivity?
if @options[:case_insensitive]
disallows_value_of(swapcased_value, @expected_message)
else
if value == swapcased_value
raise ActiveModel::CouldNotValidateCaseError.create(existing_record, @attribute)

This comment has been minimized.

@houndci-bot

houndci-bot Oct 7, 2015

Line is too long. [96/80]

def message
<<-EOT.strip
Your #{model} model has a uniqueness validation on :#{attribute} which is declared
to be case-sensitive, but the value the uniqueness matcher used, "#{value}", doesn't

This comment has been minimized.

@houndci-bot

houndci-bot Oct 7, 2015

Line is too long. [84/80]

def message
<<-EOT.strip
Your #{model} model has a uniqueness validation on :#{attribute} which is declared

This comment has been minimized.

@houndci-bot

houndci-bot Oct 7, 2015

Line is too long. [82/80]

if value == swapcased_value
raise ActiveModel::CouldNotValidateCaseError.create(
existing_record,
@attribute

This comment has been minimized.

@houndci-bot

houndci-bot Oct 7, 2015

Put a comma after the last parameter of a multiline method call.

@albertyw

This comment has been minimized.

Contributor

albertyw commented Oct 7, 2015

I've updated the error with your new text.

@mcmire - I think your original text can be copied to the wiki but I can't copy your markdown. Can you do it? (I fixed your link typo to http://github.com/thoughtbot/shoulda-matchers/wiki/error-uniques)

@mcmire

This comment has been minimized.

Collaborator

mcmire commented Oct 9, 2015

Cool. Thanks for the PR. I've merged your changes as ada9bd3. Instead of adding docs to the wiki (which I don't plan on maintaining -- really it's just user-submitted documentation), I opted to add them to the global docs. The change for that is in 80407e6.

Thanks for the discussion around this, folks!

@mcmire mcmire closed this Oct 9, 2015

@albertyw

This comment has been minimized.

Contributor

albertyw commented Oct 9, 2015

Great. Looking forward to seeing this in the next release!

@albertyw albertyw deleted the albertyw:fix-case-checking branch Oct 9, 2015

@mcmire

This comment has been minimized.

Collaborator

mcmire commented Oct 23, 2015

@albertyw @tovodeverett This is now fixed in 3.0.1.

netbsd-srcmastr pushed a commit to NetBSD/pkgsrc that referenced this pull request Sep 23, 2018

taca
devel/ruby-shoulda-matchers: update to 3.1.2
# 3.1.2

### Deprecations

* This is the **last version** that supports Rails 4.0 and 4.1 and Ruby 2.0 and 2.1.

### Bug fixes

* When the `permit` matcher was used without `#on`, the controller did not use
  `params#require`, the params object was duplicated, and the matcher did not
  recognize the `#permit` call inside the controller. This behavior happened
  because the matcher overwrote double registries with the same parameter hash
  whenever ActionController::Parameters was instantiated.

  * *Commit: [44c019]*
  * *Issue: [#899]*
  * *Pull request: [#902]*

# 3.1.1

### Bug fixes

* Some matchers make use of ActiveSupport's `in?` method, but do not include the
  file where this is defined in ActiveSupport. This causes problems with
  projects using shoulda-matchers that do not include all of ActiveSupport by
  default. To fix this, replace `in?` with Ruby's builtin `include?`.

  * *Pull request: [#879]*

* `validate_uniqueness_of` works by creating a record if it doesn't exist, and
  then testing against a new record with various attributes set that are equal
  to (or different than) corresponding attributes in the existing record. In
  3.1.0 a change was made whereby when the uniqueness matcher is given a new
  record and creates an existing record out of it, it ensures that the record is
  valid before continuing on. This created a problem because if the subject,
  before it was saved, was empty and therefore in an invalid state, it could not
  effectively be saved. While ideally this should be enforced, doing so would be
  a backward-incompatible change, so this behavior has been rolled back.
  ([#880], [#884], [#885])

  * *Commit: [45de869]*
  * *Issues: [#880], [#884], [#885]*

* Fix an issue with `validate_uniqueness_of` + `scoped_to` when used against a
  model where the attribute has multiple uniqueness validations and each
  validation has a different set of scopes. In this case, a test written for the
  first validation (and its scopes) would pass, but tests for the other
  validations (and their scopes) would not, as the matcher only considered the
  first set of scopes as the *actual* set of scopes.

  * *Commit: [28bd9a1]*
  * *Issues: [#830]*

### Improvements

* Update `validate_uniqueness_of` so that if an existing record fails to be
  created because a column is non-nullable and was not filled in, raise an
  ExistingRecordInvalid exception with details on how to fix the test.

  * *Commit: [78ccfc5]*

[#879]: thoughtbot/shoulda-matchers#879
[45de869]: thoughtbot/shoulda-matchers@45de869
[#880]: thoughtbot/shoulda-matchers#880
[#884]: thoughtbot/shoulda-matchers#884
[#885]: thoughtbot/shoulda-matchers#885
[78ccfc5]: thoughtbot/shoulda-matchers@78ccfc5
[28bd9a1]: thoughtbot/shoulda-matchers@28bd9a1
[#830]: thoughtbot/shoulda-matchers#830

# 3.1.0

### Bug fixes

* Update `validate_numericality_of` so that submatchers are applied lazily
  instead of immediately. Previously, qualifiers were order-dependent, meaning
  that if you used `strict` before you used, say, `odd`, then `strict` wouldn't
  actually apply to `odd`. Now the order that you specify qualifiers doesn't
  matter.

  * *Source: [6c67a5e]*

* Fix `allow_value` so that it does not raise an AttributeChangedValueError
  (formerly CouldNotSetAttributeError) when used against an attribute that is an
  enum in an ActiveRecord model.

  * *Source: [9e8603e]*

* Add a `ignoring_interference_by_writer` qualifier to all matchers, not just
  `allow_value`. *This is enabled by default, which means that you should never
  get a CouldNotSetAttributeError again.* (You may get some more information if
  a test fails, however.)

  * *Source: [1189934], [5532f43]*
  * *Fixes: [#786], [#799], [#801], [#804], [#817], [#841], [#849], [#872],
    [#873], and [#874]*

* Fix `validate_numericality_of` so that it does not blow up when used against
  a virtual attribute defined in an ActiveRecord model (that is, an attribute
  that is not present in the database but is defined using `attr_accessor`).

  * *Source: [#822]*

* Update `validate_numericality_of` so that it no longer raises an
  IneffectiveTestError if used against a numeric column.

  * *Source: [5ed0362]*
  * *Fixes: [#832]*

[6c67a5e]: thoughtbot/shoulda-matchers@6c67a5e
[9e8603e]: thoughtbot/shoulda-matchers@9e8603e
[1189934]: thoughtbot/shoulda-matchers@1189934
[5532f43]: thoughtbot/shoulda-matchers@5532f43
[#786]: thoughtbot/shoulda-matchers#786
[#799]: thoughtbot/shoulda-matchers#799
[#801]: thoughtbot/shoulda-matchers#801
[#804]: thoughtbot/shoulda-matchers#804
[#817]: thoughtbot/shoulda-matchers#817
[#841]: thoughtbot/shoulda-matchers#841
[#849]: thoughtbot/shoulda-matchers#849
[#872]: thoughtbot/shoulda-matchers#872
[#873]: thoughtbot/shoulda-matchers#873
[#874]: thoughtbot/shoulda-matchers#874
[#822]: thoughtbot/shoulda-matchers#822
[5ed0362]: thoughtbot/shoulda-matchers@5ed0362
[#832]: thoughtbot/shoulda-matchers#832

### Features

* Add a new qualifier, `ignoring_case_sensitivity`, to `validate_uniqueness_of`.
  This provides a way to test uniqueness of an attribute whose case is
  normalized, either in a custom writer method for that attribute, or in a
  custom `before_validation` callback.

  * *Source: [#840]*
  * *Fixes: [#836]*

[#840]: thoughtbot/shoulda-matchers#840
[#836]: thoughtbot/shoulda-matchers#836

### Improvements

* Improve failure messages and descriptions of all matchers across the board so
  that it is easier to understand what the matcher was doing when it failed.
  (You'll see a huge difference in the output of the numericality and uniqueness
  matchers in particular.)

* Matchers now raise an error if any attributes that the matcher is attempting
  to set do not exist on the model.

  * *Source: [2962112]*

* Update `validate_numericality_of` so that it doesn't always run all of the
  submatchers, but stops on the first one that fails. Since failure messages
  now contain information as to what value the matcher set on the attribute when
  it failed, this change guarantees that the correct value will be shown.

  * *Source: [8e24a6e]*

* Continue to detect if attributes change incoming values, but now instead of
  immediately seeing a CouldNotSetAttributeError, you will only be informed
  about it if the test you've written fails.

  * *Source: [1189934]*

* Add an additional check to `define_enum_for` to ensure that the column that
  underlies the enum attribute you're testing is an integer column.

  * *Source: [68dd70a]*

* Add a test for `validate_numericality_of` so that it officially supports money
  columns.

  * *Source: [a559713]*
  * *Refs: [#841]*

[2962112]: thoughtbot/shoulda-matchers@2962112
[8e24a6e]: thoughtbot/shoulda-matchers@8e24a6e
[68dd70a]: thoughtbot/shoulda-matchers@68dd70a
[a559713]: thoughtbot/shoulda-matchers@a559713

# 3.0.1

### Bug fixes

* Fix `validate_inclusion_of` + `in_array` when used against a date or datetime
  column/attribute so that it does not raise a CouldNotSetAttributeError.
  ([#783], [8fa97b4])

* Fix `validate_numericality_of` when used against a numeric column so that it
  no longer raises a CouldNotSetAttributeError if the matcher has been qualified
  in any way (`only_integer`, `greater_than`, `odd`, etc.). ([#784], [#812])

### Improvements

* `validate_uniqueness_of` now raises a NonCaseSwappableValueError if the value
  the matcher is using to test uniqueness cannot be case-swapped -- in other
  words, if it doesn't contain any alpha characters. When this is the case, the
  matcher cannot work effectively. ([#789], [ada9bd3])

[#783]: thoughtbot/shoulda-matchers#783
[8fa97b4]: thoughtbot/shoulda-matchers@8fa97b4
[#784]: thoughtbot/shoulda-matchers#784
[#789]: thoughtbot/shoulda-matchers#789
[ada9bd3]: thoughtbot/shoulda-matchers@ada9bd3
[#812]: thoughtbot/shoulda-matchers#812

# 3.0.0

### Backward-incompatible changes

* We've dropped support for Rails 3.x, Ruby 1.9.2, and Ruby 1.9.3, and RSpec 2.
  All of these have been end-of-lifed. ([a4045a1], [b7fe87a], [32c0e62])

* The gem no longer detects the test framework you're using or mixes itself into
  that framework automatically. [History][no-auto-integration-1] has
  [shown][no-auto-integration-2] that performing any kind of detection is prone
  to bugs and more complicated than it should be.

  Here are the updated instructions:

  * You no longer need to say `require: false` in your Gemfile; you can
    include the gem as normal.
  * You'll need to add the following somewhere in your `rails_helper` (for
    RSpec) or `test_helper` (for Minitest / Test::Unit):

    ``` ruby
    Shoulda::Matchers.configure do |config|
      config.integrate do |with|
        # Choose a test framework:
        with.test_framework :rspec
        with.test_framework :minitest
        with.test_framework :minitest_4
        with.test_framework :test_unit

        # Choose one or more libraries:
        with.library :active_record
        with.library :active_model
        with.library :action_controller
        # Or, choose the following (which implies all of the above):
        with.library :rails
      end
    end
    ```

  ([1900071])

* Previously, under RSpec, all of the matchers were mixed into all of the
  example groups. This created a problem because some gems, such as
  [active_model_serializers-matchers], provide matchers that share the same
  name as some of our own matchers. Now, matchers are only mixed into whichever
  example group they belong to:

    * ActiveModel and ActiveRecord matchers are available only in model example
      groups.
    * ActionController matchers are available only in controller example groups.
    * The `route` matcher is available only in routing example groups.

  ([af98a23], [8cf449b])

* There are two changes to `allow_value`:

  * The negative form of `allow_value` has been changed so that instead of
    asserting that any of the given values is an invalid value (allowing good
    values to pass through), assert that *all* values are invalid values
    (allowing good values not to pass through). This means that this test which
    formerly passed will now fail:

    ``` ruby
    expect(record).not_to allow_value('good value', *bad_values)
    ```

    ([19ce8a6])

  * `allow_value` now raises a CouldNotSetAttributeError if in setting the
    attribute, the value of the attribute from reading the attribute back is
    different from the one used to set it.

    This would happen if the writer method for that attribute has custom logic
    to ignore certain incoming values or change them in any way. Here are three
    examples we've seen:

    * You're attempting to assert that an attribute should not allow nil, yet
      the attribute's writer method contains a conditional to do nothing if
      the attribute is set to nil:

      ``` ruby
      class Foo
        include ActiveModel::Model

        attr_reader :bar

        def bar=(value)
          return if value.nil?
          @bar = value
        end
      end

      describe Foo do
        it do
          foo = Foo.new
          foo.bar = "baz"
          # This will raise a CouldNotSetAttributeError since `foo.bar` is now "123"
          expect(foo).not_to allow_value(nil).for(:bar)
        end
      end
      ```

    * You're attempting to assert that an numeric attribute should not allow a
      string that contains non-numeric characters, yet the writer method for
      that attribute strips out non-numeric characters:

      ``` ruby
      class Foo
        include ActiveModel::Model

        attr_reader :bar

        def bar=(value)
          @bar = value.gsub(/\D+/, '')
        end
      end

      describe Foo do
        it do
          foo = Foo.new
          # This will raise a CouldNotSetAttributeError since `foo.bar` is now "123"
          expect(foo).not_to allow_value("abc123").for(:bar)
        end
      end
      ```

    * You're passing a value to `allow_value` that the model typecasts into
      another value:

      ``` ruby
      describe Foo do
        # Assume that `attr` is a string
        # This will raise a CouldNotSetAttributeError since `attr` typecasts `[]` to `"[]"`
        it { should_not allow_value([]).for(:attr) }
      end
      ```

    With all of these failing examples, why are we making this change? We want
    to guard you (as the developer) from writing a test that you think acts one
    way but actually acts a different way, as this could lead to a confusing
    false positive or negative.

    If you understand the problem and wish to override this behavior so that
    you do not get a CouldNotSetAttributeError, you can add the
    `ignoring_interference_by_writer` qualifier like so. Note that this will not
    always cause the test to pass.

    ``` ruby
    it { should_not allow_value([]).for(:attr).ignoring_interference_by_writer }
    ```

    ([9d9dc4e])

* `validate_uniqueness_of` is now properly case-sensitive by default, to match
  the default behavior of the validation itself. This is a backward-incompatible
  change because this test which incorrectly passed before will now fail:

    ``` ruby
    class Product < ActiveRecord::Base
      validates_uniqueness_of :name, case_sensitive: false
    end

    describe Product do
      it { is_expected.to validate_uniqueness_of(:name) }
    end
    ```

    ([57a1922])

* `ensure_inclusion_of`, `ensure_exclusion_of`, and `ensure_length_of` have been
  removed in favor of their `validate_*` counterparts. ([55c8d09])

* `set_the_flash` and `set_session` have been changed to more closely align with
  each other:
  * `set_the_flash` has been removed in favor of `set_flash`. ([801f2c7])
  * `set_session('foo')` is no longer valid syntax, please use
    `set_session['foo']` instead. ([535fe05])
  * `set_session['key'].to(nil)` will no longer pass when the key in question
    has not been set yet. ([535fe05])

* Change `set_flash` so that `set_flash[:foo].now` is no longer valid syntax.
  You'll want to use `set_flash.now[:foo]` instead. This was changed in order to
  more closely align with how `flash.now` works when used in a controller.
  ([#755], [#752])

* Change behavior of `validate_uniqueness_of` when the matcher is not
  qualified with any scopes, but your validation is. Previously the following
  test would pass when it now fails:

  ``` ruby
  class Post < ActiveRecord::Base
    validate :slug, uniqueness: { scope: :user_id }
  end

  describe Post do
    it { should validate_uniqueness_of(:slug) }
  end
  ```

  ([6ac7b81])

[active_model_serializers-matchers]: https://github.com/adambarber/active_model_serializers-matchers
[no-auto-integration-1]: freerange/mocha@049080c
[no-auto-integration-2]: rr/rr#29
[1900071]: thoughtbot/shoulda-matchers@1900071
[b7fe87a]: thoughtbot/shoulda-matchers@b7fe87a
[a4045a1]: thoughtbot/shoulda-matchers@a4045a1
[57a1922]: thoughtbot/shoulda-matchers@57a1922
[19ce8a6]: thoughtbot/shoulda-matchers@19c38a6
[eaaa2d8]: thoughtbot/shoulda-matchers@eaaa2d8
[55c8d09]: thoughtbot/shoulda-matchers@55c8d09
[801f2c7]: thoughtbot/shoulda-matchers@801f2c7
[535fe05]: thoughtbot/shoulda-matchers@535fe05
[6ac7b81]: thoughtbot/shoulda-matchers@6ac7b81
[#755]: thoughtbot/shoulda-matchers#755
[#752]: thoughtbot/shoulda-matchers#752
[9d9dc4e]: thoughtbot/shoulda-matchers@9d9dc4e
[32c0e62]: thoughtbot/shoulda-matchers@32c0e62
[af98a23]: thoughtbot/shoulda-matchers@af98a23
[8cf449b]: thoughtbot/shoulda-matchers@8cf449b

### Bug fixes

* So far the tests for the gem have been running against only SQLite. Now they
  run against PostgreSQL, too. As a result we were able to fix some
  Postgres-related bugs, specifically around `validate_uniqueness_of`:

  * When scoped to a UUID column that ends in an "f", the matcher is able to
    generate a proper "next" value without erroring. ([#402], [#587], [#662])

  * Support scopes that are PostgreSQL array columns. Please note that this is
    only supported for Rails 4.2 and greater, as versions before this cannot
    handle array columns correctly, particularly in conjunction with the
    uniqueness validator. ([#554])

  * Fix so that when scoped to a text column and the scope is set to nil before
    running it through the matcher, the matcher does not fail. ([#521], [#607])

* Fix `define_enum_for` so that it actually tests that the attribute is present
  in the list of defined enums, as you could fool it by merely defining a class
  method that was the pluralized version of the attribute name. In the same
  vein, passing a pluralized version of the attribute name to `define_enum_for`
  would erroneously pass, and now it fails. ([#641])

* Fix `permit` so that it does not break the functionality of
  ActionController::Parameters#require. ([#648], [#675])

* Fix `validate_uniqueness_of` + `scoped_to` so that it does not raise an error
  if a record exists where the scoped attribute is nil. ([#677])

* Fix `route` matcher so if your route includes a default `format`, you can
  specify this as a symbol or string. ([#693])

* Fix `validate_uniqueness_of` so that it allows you to test against scoped
  attributes that are boolean columns. ([#457], [#694])

* Fix failure message for `validate_numericality_of` as it sometimes didn't
  provide the reason for failure. ([#699])

* Fix `shoulda/matchers/independent` so that it can be required
  independently, without having to require all of the gem. ([#746], [e0a0200])

### Features

* Add `on` qualifier to `permit`. This allows you to make an assertion that
  a restriction was placed on a slice of the `params` hash and not the entire
  `params` hash. Although we don't require you to use this qualifier, we do
  recommend it, as it's a more precise check. ([#675])

* Add `strict` qualifier to `validate_numericality_of`. ([#620])

* Add `on` qualifier to `validate_numericality_of`. ([9748869]; h/t [#356],
  [#358])

* Add `join_table` qualifier to `have_and_belong_to_many`. ([#556])

* `allow_values` is now an alias for `allow_value`. This makes more sense when
  checking against multiple values:

  ``` ruby
  it { should allow_values('this', 'and', 'that') }
  ```

  ([#692])

[9748869]: thoughtbot/shoulda-matchers@9748869
[#402]: thoughtbot/shoulda-matchers#402
[#587]: thoughtbot/shoulda-matchers#587
[#662]: thoughtbot/shoulda-matchers#662
[#554]: thoughtbot/shoulda-matchers#554
[#641]: thoughtbot/shoulda-matchers#641
[#521]: thoughtbot/shoulda-matchers#521
[#607]: thoughtbot/shoulda-matchers#607
[#648]: thoughtbot/shoulda-matchers#648
[#675]: thoughtbot/shoulda-matchers#675
[#677]: thoughtbot/shoulda-matchers#677
[#620]: thoughtbot/shoulda-matchers#620
[#693]: thoughtbot/shoulda-matchers#693
[#356]: thoughtbot/shoulda-matchers#356
[#358]: thoughtbot/shoulda-matchers#358
[#556]: thoughtbot/shoulda-matchers#556
[#457]: thoughtbot/shoulda-matchers#457
[#694]: thoughtbot/shoulda-matchers#694
[#692]: thoughtbot/shoulda-matchers#692
[#699]: thoughtbot/shoulda-matchers#699
[#746]: thoughtbot/shoulda-matchers#746
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment