Ensure methods called on association proxies respect access control. [#1083 state:resolved] [Adam Milligan, Pratik]

There are several situations it doesn't cater for, and the inconsistency isn't worth blocking 2.2.

explicitly including child associations that are also included in the parent association definition should not result in double records in the collection/double loads (#1110)
Signed-off-by: NMichael Koziarski <michael@koziarski.com>
[#1110 state:committed]

Signed-off-by: NPratik Naik <pratiknaik@gmail.com>

wrote a test showing eager loading's misbehavior (sanitizing against the wrong table) when the association has a :conditions hash
Signed-off-by: NMichael Koziarski <michael@koziarski.com>

AssociationCollection now starts transactions by calling
AssociationCollection#transaction instead of @owner.transaction or
@reflection.klass.transaction.
Signed-off-by: NMichael Koziarski <michael@koziarski.com>

[#1081 state:committed]

Signed-off-by: NPratik Naik <pratiknaik@gmail.com>

Without this change count_records will group for the count in the query and return erroneous results.
Signed-off-by: NMichael Koziarski <michael@koziarski.com>
[#937 state:committed]

Support for updating a belongs to association from the foreign key (without saving and reloading the record)
Signed-off-by: NMichael Koziarski <michael@koziarski.com>
[#142 state:committed]

Signed-off-by: NMichael Koziarski <michael@koziarski.com>

Performance: Better query for ASSOCIATION_ids. Select only ids if the association hasn't been loaded.
Signed-off-by: NJeremy Kemper <jeremy@bitsweat.net>

[#831 state:resolved]
Signed-off-by: NJeremy Kemper <jeremy@bitsweat.net>

[#348 state:resolved]
Signed-off-by: NJeremy Kemper <jeremy@bitsweat.net>

[#302 state:resolved]
Signed-off-by: NJeremy Kemper <jeremy@bitsweat.net>

* Set the association target on assignment;
* Reset target to nil on reset, rather than empty array.
Signed-off-by: NMichael Koziarski <michael@koziarski.com>
[#895 state:committed]

Signed-off-by: NMichael Koziarski <michael@koziarski.com>

This matters when the has_one is defined with an order in which case
there is an expectation that the first one will be loaded.

[#904 state:resolved]
Signed-off-by: NJeremy Kemper <jeremy@bitsweat.net>

Signed-off-by: NJeremy Kemper <jeremy@bitsweat.net>

[#903 state:resolved]
Signed-off-by: NJeremy Kemper <jeremy@bitsweat.net>

remember created records and select a random one instead of relying on sequential id values starting at 1
Signed-off-by: NMichael Koziarski <michael@koziarski.com>

Don't interpret decimals as table names in ActiveRecord::Associations::ClassMethods#references_eager_loaded_tables? [#532 state:resolved]
Signed-off-by: NPratik Naik <pratiknaik@gmail.com>

Signed-off-by: NPratik Naik <pratiknaik@gmail.com>

Signed-off-by: NPratik Naik <pratiknaik@gmail.com>

Signed-off-by: NPratik Naik <pratiknaik@gmail.com>

Signed-off-by: NPratik Naik <pratiknaik@gmail.com>

Signed-off-by: NPratik Naik <pratiknaik@gmail.com>

Signed-off-by: NMichael Koziarski <michael@koziarski.com>

Signed-off-by: NMichael Koziarski <michael@koziarski.com>

Signed-off-by: NJoshua Peek <josh@joshpeek.com>

Signed-off-by: NMichael Koziarski <michael@koziarski.com>

Signed-off-by: NPratik Naik <pratiknaik@gmail.com>

Use klass.sti_name to make sure associations take store_full_sti_class into account. [#671 state:resolved]
Signed-off-by: NPratik Naik <pratiknaik@gmail.com>

Signed-off-by: NPratik Naik <pratiknaik@gmail.com>

Signed-off-by: NPratik Naik <pratiknaik@gmail.com>