Added some assertions to enforce the atomicity of atomic.

This commit is contained in:
Aymeric Augustin 2013-03-04 23:26:31 +01:00
parent d7bc4fbc94
commit 7c46c8d5f2
14 changed files with 369 additions and 279 deletions

View file

@ -24,7 +24,7 @@ immediately committed to the database. :ref:`See below for details
.. versionchanged:: 1.6
Previous version of Django featured :ref:`a more complicated default
behavior <transactions-changes-from-1.5>`.
behavior <transactions-upgrading-from-1.5>`.
Tying transactions to HTTP requests
-----------------------------------
@ -89,7 +89,7 @@ Django provides a single API to control database transactions.
database. If this argument isn't provided, Django uses the ``"default"``
database.
``atomic`` is usable both as a decorator::
``atomic`` is usable both as a `decorator`_::
from django.db import transaction
@ -98,7 +98,7 @@ Django provides a single API to control database transactions.
# This code executes inside a transaction.
do_stuff()
and as a context manager::
and as a `context manager`_::
from django.db import transaction
@ -110,6 +110,9 @@ Django provides a single API to control database transactions.
# This code executes inside a transaction.
do_more_stuff()
.. _decorator: http://docs.python.org/glossary.html#term-decorator
.. _context manager: http://docs.python.org/glossary.html#term-context-manager
Wrapping ``atomic`` in a try/except block allows for natural handling of
integrity errors::
@ -145,189 +148,6 @@ Django provides a single API to control database transactions.
- releases or rolls back to the savepoint when exiting an inner block;
- commits or rolls back the transaction when exiting the outermost block.
.. _transaction-management-functions:
Controlling transaction management in views
===========================================
For most people, implicit request-based transactions work wonderfully. However,
if you need more fine-grained control over how transactions are managed, you can
use a set of functions in ``django.db.transaction`` to control transactions on a
per-function or per-code-block basis.
These functions, described in detail below, can be used in two different ways:
* As a decorator_ on a particular function. For example::
from django.db import transaction
@transaction.commit_on_success
def viewfunc(request):
# ...
# this code executes inside a transaction
# ...
* As a `context manager`_ around a particular block of code::
from django.db import transaction
def viewfunc(request):
# ...
# this code executes using default transaction management
# ...
with transaction.commit_on_success():
# ...
# this code executes inside a transaction
# ...
Both techniques work with all supported version of Python.
.. _decorator: http://docs.python.org/glossary.html#term-decorator
.. _context manager: http://docs.python.org/glossary.html#term-context-manager
For maximum compatibility, all of the examples below show transactions using the
decorator syntax, but all of the follow functions may be used as context
managers, too.
.. note::
Although the examples below use view functions as examples, these
decorators and context managers can be used anywhere in your code
that you need to deal with transactions.
.. _topics-db-transactions-autocommit:
.. function:: autocommit
Use the ``autocommit`` decorator to switch a view function to Django's
default commit behavior.
Example::
from django.db import transaction
@transaction.autocommit
def viewfunc(request):
....
@transaction.autocommit(using="my_other_database")
def viewfunc2(request):
....
Within ``viewfunc()``, transactions will be committed as soon as you call
``model.save()``, ``model.delete()``, or any other function that writes to
the database. ``viewfunc2()`` will have this same behavior, but for the
``"my_other_database"`` connection.
.. function:: commit_on_success
Use the ``commit_on_success`` decorator to use a single transaction for all
the work done in a function::
from django.db import transaction
@transaction.commit_on_success
def viewfunc(request):
....
@transaction.commit_on_success(using="my_other_database")
def viewfunc2(request):
....
If the function returns successfully, then Django will commit all work done
within the function at that point. If the function raises an exception,
though, Django will roll back the transaction.
.. function:: commit_manually
Use the ``commit_manually`` decorator if you need full control over
transactions. It tells Django you'll be managing the transaction on your
own.
Whether you are writing or simply reading from the database, you must
``commit()`` or ``rollback()`` explicitly or Django will raise a
:exc:`TransactionManagementError` exception. This is required when reading
from the database because ``SELECT`` statements may call functions which
modify tables, and thus it is impossible to know if any data has been
modified.
Manual transaction management looks like this::
from django.db import transaction
@transaction.commit_manually
def viewfunc(request):
...
# You can commit/rollback however and whenever you want
transaction.commit()
...
# But you've got to remember to do it yourself!
try:
...
except:
transaction.rollback()
else:
transaction.commit()
@transaction.commit_manually(using="my_other_database")
def viewfunc2(request):
....
.. _topics-db-transactions-requirements:
Requirements for transaction handling
=====================================
Django requires that every transaction that is opened is closed before the
completion of a request.
If you are using :func:`autocommit` (the default commit mode) or
:func:`commit_on_success`, this will be done for you automatically. However,
if you are manually managing transactions (using the :func:`commit_manually`
decorator), you must ensure that the transaction is either committed or rolled
back before a request is completed.
This applies to all database operations, not just write operations. Even
if your transaction only reads from the database, the transaction must
be committed or rolled back before you complete a request.
.. _managing-autocommit:
Managing autocommit
===================
.. versionadded:: 1.6
Django provides a straightforward API to manage the autocommit state of each
database connection, if you need to.
.. function:: get_autocommit(using=None)
.. function:: set_autocommit(using=None, autocommit=True)
These functions take a ``using`` argument which should be the name of a
database. If it isn't provided, Django uses the ``"default"`` database.
.. _deactivate-transaction-management:
How to globally deactivate transaction management
=================================================
Control freaks can totally disable all transaction management by setting
:setting:`TRANSACTIONS_MANAGED` to ``True`` in the Django settings file. If
you do this, Django won't enable autocommit. You'll get the regular behavior
of the underlying database library.
This requires you to commit explicitly every transaction, even those started
by Django or by third-party libraries. Thus, this is best used in situations
where you want to run your own transaction-controlling middleware or do
something really strange.
In almost all situations, you'll be better off using the default behavior, or
the transaction middleware, and only modify selected functions as needed.
.. _topics-db-transactions-savepoints:
Savepoints
@ -339,13 +159,19 @@ available with the SQLite (≥ 3.6.8), PostgreSQL, Oracle and MySQL (when using
the InnoDB storage engine) backends. Other backends provide the savepoint
functions, but they're empty operations -- they don't actually do anything.
Savepoints aren't especially useful if you are using the default
``autocommit`` behavior of Django. However, if you are using
``commit_on_success`` or ``commit_manually``, each open transaction will build
up a series of database operations, awaiting a commit or rollback. If you
issue a rollback, the entire transaction is rolled back. Savepoints provide
the ability to perform a fine-grained rollback, rather than the full rollback
that would be performed by ``transaction.rollback()``.
Savepoints aren't especially useful if you are using autocommit, the default
behavior of Django. However, once you open a transaction with :func:`atomic`,
you build up a series of database operations awaiting a commit or rollback. If
you issue a rollback, the entire transaction is rolled back. Savepoints
provide the ability to perform a fine-grained rollback, rather than the full
rollback that would be performed by ``transaction.rollback()``.
.. versionchanged:: 1.6
When the :func:`atomic` decorator is nested, it creates a savepoint to allow
partial commit or rollback. You're strongly encouraged to use :func:`atomic`
rather than the functions described below, but they're still part of the
public API, and there's no plan to deprecate them.
Each of these functions takes a ``using`` argument which should be the name of
a database for which the behavior applies. If no ``using`` argument is
@ -374,15 +200,17 @@ The following example demonstrates the use of savepoints::
from django.db import transaction
@transaction.commit_manually
# open a transaction
@transaction.atomic
def viewfunc(request):
a.save()
# open transaction now contains a.save()
# transaction now contains a.save()
sid = transaction.savepoint()
b.save()
# open transaction now contains a.save() and b.save()
# transaction now contains a.save() and b.save()
if want_to_keep_b:
transaction.savepoint_commit(sid)
@ -391,7 +219,82 @@ The following example demonstrates the use of savepoints::
transaction.savepoint_rollback(sid)
# open transaction now contains only a.save()
transaction.commit()
Autocommit
==========
.. _autocommit-details:
Why Django uses autocommit
--------------------------
In the SQL standards, each SQL query starts a transaction, unless one is
already in progress. Such transactions must then be committed or rolled back.
This isn't always convenient for application developers. To alleviate this
problem, most databases provide an autocommit mode. When autocommit is turned
on, each SQL query is wrapped in its own transaction. In other words, the
transaction is not only automatically started, but also automatically
committed.
:pep:`249`, the Python Database API Specification v2.0, requires autocommit to
be initially turned off. Django overrides this default and turns autocommit
on.
To avoid this, you can :ref:`deactivate the transaction management
<deactivate-transaction-management>`, but it isn't recommended.
.. versionchanged:: 1.6
Before Django 1.6, autocommit was turned off, and it was emulated by
forcing a commit after write operations in the ORM.
.. warning::
If you're using the database API directly — for instance, you're running
SQL queries with ``cursor.execute()`` — be aware that autocommit is on,
and consider wrapping your operations in a transaction, with
:func:`atomic`, to ensure consistency.
.. _managing-autocommit:
Managing autocommit
-------------------
.. versionadded:: 1.6
Django provides a straightforward API to manage the autocommit state of each
database connection, if you need to.
.. function:: get_autocommit(using=None)
.. function:: set_autocommit(using=None, autocommit=True)
These functions take a ``using`` argument which should be the name of a
database. If it isn't provided, Django uses the ``"default"`` database.
Autocommit is initially turned on. If you turn it off, it's your
responsibility to restore it.
:func:`atomic` requires autocommit to be turned on; it will raise an exception
if autocommit is off. Django will also refuse to turn autocommit off when an
:func:`atomic` block is active, because that would break atomicity.
.. _deactivate-transaction-management:
Deactivating transaction management
-----------------------------------
Control freaks can totally disable all transaction management by setting
:setting:`TRANSACTIONS_MANAGED` to ``True`` in the Django settings file. If
you do this, Django won't enable autocommit. You'll get the regular behavior
of the underlying database library.
This requires you to commit explicitly every transaction, even those started
by Django or by third-party libraries. Thus, this is best used in situations
where you want to run your own transaction-controlling middleware or do
something really strange.
In almost all situations, you'll be better off using the default behavior, or
the transaction middleware, and only modify selected functions as needed.
Database-specific notes
=======================
@ -477,45 +380,57 @@ transaction. For example::
In this example, ``a.save()`` will not be undone in the case where
``b.save()`` raises an exception.
Under the hood
==============
.. _transactions-upgrading-from-1.5:
.. _autocommit-details:
Changes from Django 1.5 and earlier
===================================
Details on autocommit
---------------------
The features described below were deprecated in Django 1.6 and will be removed
in Django 1.8. They're documented in order to ease the migration to the new
transaction management APIs.
In the SQL standards, each SQL query starts a transaction, unless one is
already in progress. Such transactions must then be committed or rolled back.
Legacy APIs
-----------
This isn't always convenient for application developers. To alleviate this
problem, most databases provide an autocommit mode. When autocommit is turned
on, each SQL query is wrapped in its own transaction. In other words, the
transaction is not only automatically started, but also automatically
committed.
The following functions, defined in ``django.db.transaction``, provided a way
to control transactions on a per-function or per-code-block basis. They could
be used as decorators or as context managers, and they accepted a ``using``
argument, exactly like :func:`atomic`.
:pep:`249`, the Python Database API Specification v2.0, requires autocommit to
be initially turned off. Django overrides this default and turns autocommit
on.
.. function:: autocommit
To avoid this, you can :ref:`deactivate the transaction management
<deactivate-transaction-management>`, but it isn't recommended.
Enable Django's default autocommit behavior.
.. versionchanged:: 1.6
Before Django 1.6, autocommit was turned off, and it was emulated by
forcing a commit after write operations in the ORM.
Transactions will be committed as soon as you call ``model.save()``,
``model.delete()``, or any other function that writes to the database.
.. warning::
.. function:: commit_on_success
If you're using the database API directly — for instance, you're running
SQL queries with ``cursor.execute()`` — be aware that autocommit is on,
and consider wrapping your operations in a transaction to ensure
consistency.
Use a single transaction for all the work done in a function.
If the function returns successfully, then Django will commit all work done
within the function at that point. If the function raises an exception,
though, Django will roll back the transaction.
.. function:: commit_manually
Tells Django you'll be managing the transaction on your own.
Whether you are writing or simply reading from the database, you must
``commit()`` or ``rollback()`` explicitly or Django will raise a
:exc:`TransactionManagementError` exception. This is required when reading
from the database because ``SELECT`` statements may call functions which
modify tables, and thus it is impossible to know if any data has been
modified.
.. _transaction-states:
Transaction management states
-----------------------------
Transaction states
------------------
The three functions described above relied on a concept called "transaction
states". This mechanisme was deprecated in Django 1.6, but it's still
available until Django 1.8..
At any time, each database connection is in one of these two states:
@ -529,35 +444,80 @@ Django starts in auto mode. ``TransactionMiddleware``,
Internally, Django keeps a stack of states. Activations and deactivations must
be balanced.
For example, at the beginning of each HTTP request, ``TransactionMiddleware``
switches to managed mode; at the end of the request, it commits or rollbacks,
For example, ``commit_on_success`` switches to managed mode when entering the
block of code it controls; when exiting the block, it commits or rollbacks,
and switches back to auto mode.
.. admonition:: Nesting decorators / context managers
So :func:`commit_on_success` really has two effects: it changes the
transaction state and it defines an transaction block. Nesting will give the
expected results in terms of transaction state, but not in terms of
transaction semantics. Most often, the inner block will commit, breaking the
atomicity of the outer block.
:func:`commit_on_success` has two effects: it changes the transaction
state, and defines an atomic transaction block.
:func:`autocommit` and :func:`commit_manually` have similar limitations.
Nesting with :func:`autocommit` and :func:`commit_manually` will give the
expected results in terms of transaction state, but not in terms of
transaction semantics. Most often, the inner block will commit, breaking
the atomicity of the outer block.
API changes
-----------
Django currently doesn't provide any APIs to create transactions in auto mode.
Managing transactions
~~~~~~~~~~~~~~~~~~~~~
.. _transactions-changes-from-1.5:
Starting with Django 1.6, :func:`atomic` is the only supported API for
defining a transaction. Unlike the deprecated APIs, it's nestable and always
guarantees atomicity.
Changes from Django 1.5 and earlier
===================================
In most cases, it will be a drop-in replacement for :func:`commit_on_success`.
During the deprecation period, it's possible to use :func:`atomic` within
:func:`autocommit`, :func:`commit_on_success` or :func:`commit_manually`.
However, the reverse is forbidden, because nesting the old decorators /
context managers breaks atomicity.
If you enter :func:`atomic` while you're in managed mode, it will trigger a
commit to start from a clean slate.
Managing autocommit
~~~~~~~~~~~~~~~~~~~
Django 1.6 introduces an explicit :ref:`API for mananging autocommit
<managing-autocommit>`.
To disable autocommit temporarily, instead of::
with transaction.commit_manually():
# do stuff
you should now use::
transaction.set_autocommit(autocommit=False)
try:
# do stuff
finally:
transaction.set_autocommit(autocommit=True)
To enable autocommit temporarily, instead of::
with transaction.autocommit():
# do stuff
you should now use::
transaction.set_autocommit(autocommit=True)
try:
# do stuff
finally:
transaction.set_autocommit(autocommit=False)
Backwards incompatibilities
---------------------------
Since version 1.6, Django uses database-level autocommit in auto mode.
Previously, it implemented application-level autocommit by triggering a commit
after each ORM write.
As a consequence, each database query (for instance, an
ORM read) started a transaction that lasted until the next ORM write. Such
"automatic transactions" no longer exist in Django 1.6.
As a consequence, each database query (for instance, an ORM read) started a
transaction that lasted until the next ORM write. Such "automatic
transactions" no longer exist in Django 1.6.
There are four known scenarios where this is backwards-incompatible.
@ -565,7 +525,7 @@ Note that managed mode isn't affected at all. This section assumes auto mode.
See the :ref:`description of modes <transaction-states>` above.
Sequences of custom SQL queries
-------------------------------
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
If you're executing several :ref:`custom SQL queries <executing-custom-sql>`
in a row, each one now runs in its own transaction, instead of sharing the
@ -577,20 +537,20 @@ usually followed by a call to ``transaction.commit_unless_managed``, which
isn't necessary any more and should be removed.
Select for update
-----------------
~~~~~~~~~~~~~~~~~
If you were relying on "automatic transactions" to provide locking between
:meth:`~django.db.models.query.QuerySet.select_for_update` and a subsequent
write operation — an extremely fragile design, but nonetheless possible — you
must wrap the relevant code in :func:`commit_on_success`.
must wrap the relevant code in :func:`atomic`.
Using a high isolation level
----------------------------
~~~~~~~~~~~~~~~~~~~~~~~~~~~~
If you were using the "repeatable read" isolation level or higher, and if you
relied on "automatic transactions" to guarantee consistency between successive
reads, the new behavior is backwards-incompatible. To maintain consistency,
you must wrap such sequences in :func:`commit_on_success`.
reads, the new behavior might be backwards-incompatible. To enforce
consistency, you must wrap such sequences in :func:`atomic`.
MySQL defaults to "repeatable read" and SQLite to "serializable"; they may be
affected by this problem.
@ -602,10 +562,9 @@ PostgreSQL and Oracle default to "read committed" and aren't affected, unless
you changed the isolation level.
Using unsupported database features
-----------------------------------
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
With triggers, views, or functions, it's possible to make ORM reads result in
database modifications. Django 1.5 and earlier doesn't deal with this case and
it's theoretically possible to observe a different behavior after upgrading to
Django 1.6 or later. In doubt, use :func:`commit_on_success` to enforce
integrity.
Django 1.6 or later. In doubt, use :func:`atomic` to enforce integrity.