File operations always raise a ENOENT error when a file doesn't exist.
Checking the file exists before the operation adds a race condition
condition where the file could be removed between operations. As the
operation already raises an error on a missing file, avoid this race and
avoid checking the file exists twice. Instead only check a file exists
by catching the ENOENT error.
This also fixes a possible data eviction race condition between
setting and getting a key. Another thread could remove the key
before get_and_set() accesses it again. In this case, now the
default value will be returned instead of None.
The original intent in refs #20536 was to use the highest protocol.
Calling zlib.compress() with a compression level of -1 seems to
fall back to the default level of 6.
The new signature enables better support for routing RunPython and
RunSQL operations, especially w.r.t. reusable and third-party apps.
This commit also takes advantage of the deprecation cycle for the old
signature to remove the backward incompatibility introduced in #22583;
RunPython and RunSQL won't call allow_migrate() when when the router
has the old signature.
Thanks Aymeric Augustin and Tim Graham for helping shape up the patch.
Refs 22583.
Default Memcached configuration allows for a maximum object of 1MB and
will fail to set the key if it is too large. The key will be deleted from
memcached if it fails to be set. This is needed to avoid an issue with
cache_db session backend using the old value stored in memcached, instead
of the newer value stored in the database.
This feature allows the default `TIMEOUT` Cache argument to be set to `None`,
so that cache instances can set a non-expiring key as the default,
instead of using the default value of 5 minutes.
Previously, this was possible only by passing `None` as an argument to
the set() method of objects of type `BaseCache` (and subtypes).
This commit touchs various parts of the code base and test framework. Any
found usage of opening a cursor for the sake of initializing a connection
has been replaced with 'ensure_connection()'.
Thanks Curtis Malony and Florian Apolloner.
Squashed commit of the following:
commit 3380495e93
Author: Aymeric Augustin <aymeric.augustin@m4x.org>
Date: Sat Nov 23 14:18:07 2013 +0100
Looked up the template_fragments cache at runtime.
commit 905a74f52b
Author: Aymeric Augustin <aymeric.augustin@m4x.org>
Date: Sat Nov 23 14:19:48 2013 +0100
Removed all uses of create_cache.
Refactored the cache tests significantly.
Made it safe to override the CACHES setting.
commit 35e289fe92
Author: Aymeric Augustin <aymeric.augustin@m4x.org>
Date: Sat Nov 23 12:23:57 2013 +0100
Removed create_cache function.
commit 8e274f747a
Author: Aymeric Augustin <aymeric.augustin@m4x.org>
Date: Sat Nov 23 12:04:52 2013 +0100
Updated docs to describe a simplified cache backend API.
commit ee7eb0f73e
Author: Curtis Maloney <curtis@tinbrain.net>
Date: Sat Oct 19 09:49:24 2013 +1100
Fixed#21012 -- Thread-local caches, like databases.
* Safer for use in multiprocess environments
* Better random culling
* Cache files use less disk space
* Safer delete behavior
Also fixed#15806, fixed#15825.
The precision of time.time() is OS specific and it is possible for the
resolution to be low enough to allow reading a cache key previously set
with a timeout of 0.
DatabaseCache uses raw cursors to bypass the ORM. This prevents it from
being used by database backends that require special handling of datetime
values.
There is no easy way to test this, so no tests added.
Thanks Anssi for haggling until I implemented this.
This change alleviates the need for atomic_if_autocommit. When
autocommit is disabled for a database, atomic will simply create and
release savepoints, and not commit anything. This honors the contract of
not doing any transaction management.
This change also makes the hack to allow using atomic within the legacy
transaction management redundant.
None of the above will work with SQLite, because of a flaw in the design
of the sqlite3 library. This is a known limitation that cannot be lifted
without unacceptable side effects eg. triggering arbitrary commits.
Since "unless managed" now means "if database-level autocommit",
committing or rolling back doesn't have any effect.
Restored transactional integrity in a few places that relied on
automatically-started transactions with a transitory API.
There was a problem caused by Postgres 9.0+ having bytea_output default
value of 'hex' and cache backend inserting the content as 'bytes' into
a column of type TEXT. Fixed by converting the bytes value to a string
before insert.
* Renamed smart_unicode to smart_text (but kept the old name under
Python 2 for backwards compatibility).
* Renamed smart_str to smart_bytes.
* Re-introduced smart_str as an alias for smart_text under Python 3
and smart_bytes under Python 2 (which is backwards compatible).
Thus smart_str always returns a str objects.
* Used the new smart_str in a few places where both Python 2 and 3
want a str.
Also removed the hasattr check when firing request_finished signal for
caches with a 'close' method. Should be safe to call `cache.close`
everywhere now
Ville Skyttä
Jon Dufresne
Tim Graham
Przemysław Suliga
userimack
Matthew Madurski
Emre Yilmaz
Jaap Roes
Andrew Artajos
Tom Christie
Matthew Crowson
Maxime Lorant
Karol Duleba
Aymeric Augustin
Piotr Jakimiak
Berker Peksag
Loic Bistuer
Michael Manfre
Thomas Chaumeny
Jaap Roes
Vincent-Vega
Malcolm Box
Claude Paroz
zedr
Alex Gaynor
Florian Apolloner
Curtis Maloney
coagulant
Gregor MacGregor
Simon Charette
ersran9
Jacob Burch
Tomek Paczkowski
Bas Peschier
George Song
Anssi Kääriäinen
Alex Gaynor
Mike Grouchy
Aymeric Augustin
Jannis Leidel