Skip to content
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

Support case-insensitive uniqueness validations under MySQL #39

Open
wants to merge 3 commits into
base: master
Choose a base branch
from

Conversation

dmeranda
Copy link

This is a proposed fix for issue #38

I'm not sure how to proceed writing a test spec for this. It should only be run with against a MySQL database, and furthermore may require creating a table using raw SQL (execute) rather than as a rails schema migration/load; because a proper test will rely on the MySQL COLLATION type qualifier which is not exposed by the schema loader/dumper (though it is a read-only property of the Column type).

The example code presented in the issue would probably serve as a good basis for a test.

In MySQL individual columns can be case-insensitive, according to
their collation type qualifier, without regard to any index.
Rails already provides a case_sensitive? method on columns via
the AbstractMysqlAdapter module.
…lation.

When determining whether a uniqueness validation should be case insensitive,
we only want to consider the column's own reported case-sensitivity if it
is a true property of the column, i.e., due to its collation/character set.

The schema_plus_columns gem monkey patches in its own version of
case_sensitive?, for PostgreSQL, that determines this based upon the
properties of ANY indexes in which the column participates. That
method does NOT correctly take into account the index scope.  Since in
PostgreSQL it is possible to have a column participate in two indexes,
one being case-sensitive and another being case-insensitive, it is
important to always consider the scope.  This does not matter with MySQL.
@coveralls
Copy link

Coverage Status

Coverage increased (+0.01%) to 99.275% when pulling 89f54f0 on dmeranda:mysql-case-insensitive into b3e8be2 on SchemaPlus:master.

Old text implied case-insensitivity is only supported with that gem.
But now MySQL case-insensitive validations will work without
any extra gems.
@coveralls
Copy link

Coverage Status

Coverage increased (+0.01%) to 99.275% when pulling ac3ed28 on dmeranda:mysql-case-insensitive into b3e8be2 on SchemaPlus:master.

@ronen
Copy link
Member

ronen commented Apr 20, 2016

Thanks for the PR. Good catch regarding :collation. Arguably, schema_plus_columns case_sensitive? method is wrong or misleading in its logic. Maybe it should take an optional scope argument. I'll open an issue over there.

I'm not sure how to proceed writing a test spec for this. It should only be run with against a MySQL database,

Schema_dev adds the ability for you to specify, e.g. :mysql => :only or :mysql => :skip as an option on any rspec context or test. See for example validations_spec.rb#L112 or schema_plus_views/spec/named_schemas_spec.rb#L63

and furthermore may require creating a table using raw SQL (execute) rather than as a rails schema migration/load

schema_plus_views/spec/named_schemas_spec.rb also has examples of creating schemas via connection.execute, which you ought to be able to mimic -- and this case should be simpler than that since it'd be mysql only.

If you can fit it into validations_spec.rb i guess that'd be best, otherwise it could be in a separate file. I actually don't like validations_spec's setup of having one big ugly schema shared by all tests; each test or context should define a small schema containing just the bits that are needed for that specific test. This big-ugly-schema style is legacy code that I've been gradually moving away from in the schema plus family. So you shouldn't feel the need to shoehorn the collation stuff into that big schema, just define something specific to the new the case(s) you're testing.

the MySQL COLLATION type qualifier which is not exposed by the schema loader/dumper (though it is a read-only property of the Column type).

That seems like a natural thing for schema_plus_columns to add! I'll open up an issue there for that too.

Thanks again for taking the time with this!

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

3 participants