Rom

[changed] Cleaned up imports and __all__ that was put off after the refactor
[changed] Entities whose underlying data has changed will raise either the
DataRaceError or EntityDeletedError exception, depending on whether the
underlying data was changed, or the full entity was deleted. To force save
regardless of underlying changes, you can use entity.save(force=True), or
the various session methods that also take the force=True argument. Thanks
to Github user https://github.com/jmmills for the bug report. Note that
these exceptions will only be raised when Lua writes are enabled
[added] Two new keygen functions for string indexes - IDENTITY and
IDENTITY_CI, that allow you to search for exact strings and
case-insensitive exact strings. Initial request for keygen thanks to
Github user https://github.com/peter-conalgo
[changed] The CASE_INSENSITIVE keygen has been renamed to SIMPLE_CI to
match the naming of IDENTITY/IDENTITY_CI. The old name is still available,
but will be deprecated in the future
[added] warning of future deprecation of disabled Lua writes - we're 3 years
into viable Lua support in Redis, and there is no good reason not to
enable it

[added] empty base methods for handling commit hooks:
_before_insert(), _after_insert(),
_before_update(), _after_update(),
_before_delete(), _after_delete() - you should override them in your Model
subclass to override their default (do nothing) behavior. Feature request
thanks to Github user https://github.com/peter-conalgo
[added] when you need to query a ManyToOne or OneToOne column for references
that are not None/NULL, you can now use Model.query.filter(col=NOT_NULL)
and get the answer you are looking for. Feature request thanks to Github
user https://github.com/peter-conalgo

[changed] made Model.query.all() (a bare query for all entities) behave like
list(Model.query), thanks to feature request by
https://github.com/peter-conalgo
[changed] made Model.query.first() (a bare query for the 'first' entity)
behave like `for i in Model.query: return i`, thanks to feature request by
https://github.com/peter-conalgo
[changed] made empty queries against the Model faster by using an index on the
the PrimaryKey() column if it is available, or HSCAN + a temporary set
when it is not available and you are using Redis 2.8+
[changed] pulled the Query, Model, and _ModelMetaclass objects out of the
__init__.py file and moved them into their own modules - they were getting
too big
[fixed] added missing OneToOne entry in the __all__ exports in __init__.py

[fixed] index cleanup with custom namespaces have been broken since their
introduction in 0.29.4, now fixed.
[added] the ability to get a total count of all entities with a bare
Model.query.count(), thanks to feature request by Github user
https://github.com/peter-conalgo
[added] the ability to iterate over all entities without requiring an explicit
index or filtering with a query with Model.query.iter_results(), thanks to
feature request by Github user https://github.com/peter-conalgo
[added] base iterator on Model.query so if you are happy with the defaults,
you can use `for i in Model.query:` instead of
`for i in Model.query.iter_results():`, thanks to feature request by
https://github.com/peter-conalgo
[fixed] `Model.query.iter_results()` on non-empty queries with a limit clause
set could sometimes return more results than were requested. This is now
fixed.
[fixed/changed] Model.query.count() will now honor limit clauses.

[fixed] an issue when attempting to fetch a broken entity. Bug report and
initial patch thanks to github user https://github.com/huangwei1024
[added] QueryError exceptions when trying to use .filter(), .startswith(),
.endswith(), .like(), or .order_by() without the appropriate indexes.

[added] a _connection property to all Models to be able to fetch the
connection that would be used to read or write data for the model. Thanks
to https://github.com/jmmills for the feature request