67cccf3e8a
[entropy.client] lazy load ClientSystemSettingsPlugin
2013-12-12 17:13:19 +01:00
8a7afe97e8
[entropy.client] use ClientSettings() whenever needed
...
This commit also drops the sys_settings_client_plugin_id attribute
2013-12-12 16:49:01 +01:00
56f0ed4fd5
[entropy.client] Client: move object attributes to @property to enable lazy loading
2013-12-12 15:05:02 +01:00
3a7269aa5c
[entropy.db] EntropyBaseRepository: turn lock methods into no-op if direct mode
...
Now that the repository lock is reentrant, it's good to have the
methods take into account direct mode as well. In direct mode,
we explicitly don't want to deal with any kind of locking, because
we accept to manipulate stale data. In order to hide locking code
from the outside and have it transparently managed inside entropy.*
methods, we must respect requests made in direct mode.
2013-12-12 14:25:44 +01:00
412759e20f
Release Entropy 251
2013-12-12 12:38:28 +01:00
db8bd0fb40
[entropy.client.package] multifetch: ensure that path exists before calling _match_checksum
2013-12-12 12:36:45 +01:00
4b8b3b1af7
[entropy.client.package] multifetch: use the same path for both calls
2013-12-12 12:36:03 +01:00
864e3ecf21
[entropy.client.package] multifetch: fix exit status in case of download failures
2013-12-12 12:30:25 +01:00
36b634aea7
[entropy.output] _print_prio: use nice unicode header chars :-)
2013-12-12 11:56:04 +01:00
e21d564669
Release Entropy 250
2013-12-12 10:50:15 +01:00
d04f67c96f
[entropy.db.sqlite] fix typo in argument name
2013-12-12 10:41:21 +01:00
023e8e1b72
[entropy.db] EntropyBaseRepository: make possible to permanently enable direct mode
2013-12-12 10:21:27 +01:00
af2ce34d57
Release Entropy 249
2013-12-11 22:44:44 +01:00
856184b530
[entropy.client.db] do not hide the exception traceback
2013-12-11 22:43:02 +01:00
0243ed70b1
[entropy.db.sql] rewrite type checking in insertDependencies()
...
metadata could be of list type (and not just tuple) in case of
data coming from websrv.
2013-12-11 22:42:14 +01:00
eada088f56
[entropy.client.db] do not catch KeyError for update() but just during object init
2013-12-11 22:19:40 +01:00
43f4bd5ee3
Release Entropy 248
2013-12-11 16:39:02 +01:00
68fe99d33e
[entropy.db.sqlite] acquire an exclusive lock if schema updates are needed
...
_maybeDatabaseSchemaUpdates is only called inside the costructor,
at that time, it's not possible to acquire the lock elsewhere
because the object doesn't exist yet.
2013-12-11 07:49:04 +01:00
061b02bebc
[entropy.db.sqlite] move indexing setup back to object constructor
...
This has nothing to do with schema update
2013-12-11 07:43:50 +01:00
de88d6cc73
[entropy.client.repository] wrap Repository.sync() around a ResourceLock
2013-12-10 22:22:41 +01:00
82d6d37bcf
[entropy.db.sqlite] move schema update code away from __init__
2013-12-10 21:58:19 +01:00
7808ad6b92
[entropy.db.sqlite] migrate locking code to entropy.locks.ResourceLock
...
The new EntropySQLiteRepository uses ResourceLock, and gains support
for reentrancy, anti-deadlock safety measures (only for nested calls),
unification of memory and file repositories locking code (the semantics
was already the same).
2013-12-10 21:42:34 +01:00
401b9cbeb3
[entropy.client.loaders] Security: pass args and kwargs down to the object constructor
2013-12-10 16:42:34 +01:00
498674828d
[entropy.tools] get_hash_from_md5_file: do not hide IOError exceptions
2013-12-10 16:39:55 +01:00
fc597d04c5
[entropy.security] refactor System, add internal locking support, improve speed
2013-12-10 16:39:17 +01:00
b99def9128
[entropy.locks] add exclusive() and shared() context managers
2013-12-10 11:22:42 +01:00
47e9b6ee0e
[entropy.locks] rename _GenericResourceLock to ResourceLock
2013-12-10 05:39:59 +01:00
94ab49fd0b
[tests] use const_mk{s,d}temp instead of tmpfile ones, default TMPDIR to /var/tmp
2013-12-09 14:36:34 +01:00
6a2d45df47
[entropy.locks] set recursed to False only when the reference counter is 0
2013-12-09 14:24:37 +01:00
d080db0d3b
[tests] locks: fix test class name
2013-12-09 14:21:05 +01:00
bf3a8b79f9
[entropy.locks] use TLS for reentrancy safety checks, update tests
2013-12-09 14:19:21 +01:00
7e30b65744
[entropy.locks] _file_lock_create: beautify, open O_CREAT | O_RDONLY if shared mode
2013-12-08 16:33:29 +01:00
ebf63e6060
[entropy.locks] fix shared/exclusive locking mix due to reentrancy, add tests
2013-12-08 16:15:39 +01:00
97f18338ba
[entropy.locks] move acquire_lock(), release_lock() from entropy.tools
2013-12-08 14:54:42 +01:00
d0f11728c8
[entropy.security] remove unlocked_sync(), locking happens externally nowadays
2013-12-08 14:46:05 +01:00
358c6cc80c
[entropy.client.repository] drop unlocked_sync(), locking happens externally nowadays
2013-12-08 14:42:23 +01:00
1917771b54
[entropy.locks] move Entropy Resources Lock code to a separate module
2013-12-08 14:37:57 +01:00
b1ed42d80a
[entropy.db.skel] drop __atomMatchValidateCache
...
SHA1 is used nowadays and all the caching bugs should be fixed
rather than being hidden.
2013-12-08 06:28:08 +01:00
1688ce53ac
[entropy.client] clean_downloaded_packages, add skip_available_packages keyword arg
...
skip_available_packages can be used to filter out the removal of
package tarballs still belonging to available packages. This makes
possible to just use a shared Entropy Resources lock.
2013-12-07 20:30:07 +01:00
16025afc75
[entropy.client.package] move "QA:" to translatable strings
2013-12-07 20:30:07 +01:00
e75b95a1b3
[entropy.client.package] workout_subdir: do not rmtree if symlink expected but directory found
2013-12-07 20:30:06 +01:00
e6cca25473
[entropy.client.package] fail (rather than rmtree) when a file is expected but directory is found
2013-12-07 20:30:06 +01:00
0d486bc470
[entropy.client.package] if a directory is a file, do not unlink() but rename()
2013-12-07 20:30:06 +01:00
a0d4711cdc
[entropy.client.package.actions] remove: commit after insertSpmUid()
2013-12-07 20:30:06 +01:00
6d4f9e3089
[entropy.client.package.actions] install: commit after insertSpmUid()
2013-12-07 20:30:06 +01:00
b26842e826
[entropy.db] direct: wrap yield around a try/finally block
2013-12-07 20:30:06 +01:00
3545f2b619
[entropy.db] add direct access support to repository
...
In latency sensitive code paths, the performance penality caused
by file lock contention and memory cache invalidation is too high.
This problem happens in Rigo, which is extremely latency sensitive.
Since we don't want to crap on the user, a way to solve this is
letting API consumers skip the memory cache and read data directly
from the database store. The trade off is that data may be stale,
incomplete, or invalid, but as long as the consumer is aware of this,
that's fine.
2013-12-07 20:30:06 +01:00
f8d0511506
[entropy.client.package.actions] multifetch: fix pre_download_hook for edeltas
2013-12-07 20:30:06 +01:00
65f5dce076
[entropy.client.package] add sys import
2013-12-07 20:30:06 +01:00
a8c98a5081
[entropy.client.package.actions] action: cope with unavailability of file locking
...
Some filesystems, especially old network file systems, don't support
file locking. Try not to explode if this is the case but rather nag
the user with scary error messages.
2013-12-07 20:30:06 +01:00
Fabio Erculiani