According to mapping drivers the discriminator values can always be
converted to strings so it's safe to assume that we can actually do a
strict comparison during hydration.
Hydrator was ignoring data from subclasses when using multiple
inheritance levels. With this patch it will now use the discriminator
values from all subclasses of the class being hydrated.
The `v2.5.x` series of the ORM allowed to have DELETE DQLs without using
an alias, even though it didn't follow the grammar rules of the parser.
We fixed that issue on `v2.6.0` however that was a BC-breaking change
and lots of people were relying on this faulty behaviour.
This workaround fixes the BC-break, without even trying to be elegant.
In `v2.7.0.` we should raise a deprecation notice to notify people that
we'll drop that "feature" in `v3.0`.
Now using PHP to calculate the expected date manipulation, keeping a day
as delta since PHP resets the hour when performing operations with
days/weeks/months/years.
February is a wonderful month, isn't it?
When memcached extension is loaded Doctrine\ORM\Tools\Setup.php calls to setMemcache method. The MemcachedCache class has the setMemcached method instead. Changed this call in Setup to setMemcached and $memcache to $memcached to keep the name like the extension
SchemaTool has custom logic for creating the primary key of a
joined-table inherited entity. This logic overlooked association maps
as a possible source for identity columns, resulting in a fatal error
when fetching the primary key list for child entities.
Removed any custom logic for generating primary keys for root entities
in joined-table inheritance, deferring to the common logic used for
other entities.
Also adjusted the child entity logic, scanning association maps for
identity columns, and including the column as appropriate. It also
ensures that the primary key columns are in the correct order.
This will make it easier to contribute. It will also make reviews safer
since you will be able to tell if a PR is reponsible for a broken build
or not. Catching errors early is still possible by having a look at
allowed failures.
This will provide the same look for all of the relations: one-to-many, many-to-one, one-to-one, many-to-many.
It helps during auto-completion of XML code when creating XML schema for an entity.
The order is as follows: field, target-entity, mapped-by, inversed-by, indexed-by, fetch, orphan-removal.
