Follows-up 8f5cd11d82.
The old getLocalFileReferences() method is no longer used anywhere.
Remove it and rename getAllLocalFileReferences back to it.
Change-Id: I864258aad128ba9b54464c7bc854543f2937f977
Available as of PHP 5.5 and more idomatic. Foo::class (explicit),
self::class (defined), and static::class (late bound).
Change-Id: I66937f32095a4e4ecde94ca20a935a3c3efc9cee
I searched for /\$(\S+) = (.+?\(.*?\);)\n.*?\$\1\[/, ignored
everything involving isset(), unset() or array assigments, then
skimmed through the remaining results and changed things where they
made sense. These changes were not automated, so please review them.
Change-Id: Ib37b4c66fc57648470f151ad412210b3629c2538
Two things were wrong here:
1. SessionManager shouldn't override the configured logger of the
underlying store.
2. CachedBagOStuff shouldn't override the configured logger of the
backend store on construction.
Change-Id: I24ed5a81d5b238a42934a7890dd8d0a9f9bb3b6f
* Fix errors spotted by new release
* Introduce "composer fix", which uses phpcbf to automatically fix some
errors spotted by phpcs.
* Drop $PHPCS_ARGS variable that didn't work on Windows, and add -s flag
* Remove rules from phpcs.xml that are now in MW-CS ruleset.
Change-Id: I13e2155695918c918b67497ac65b85a03897095e
Ie161e0f was done in a hurry, and so didn't do things in the best ways.
This introduces a new "CachedBagOStuff" that transparently handles all
the logic that had been copy-pasted all over in Ie161e0f.
The differences between CachedBagOStuff and MultiWriteBagOStuff are:
* CachedBagOStuff supports only one "backend".
* There's a flag for writes to only go to the in-memory cache.
* The in-memory cache is always updated.
* Locks go to the backend cache (with MultiWriteBagOStuff, it would wind
up going to the HashBagOStuff used for the in-memory cache).
Change-Id: Iea494729bd2e8c6c5ab8facf4c241232e31e8215
Keep in CSSMin as-is for back-compat and to ensure library remains
independent of MediaWiki.
Moved down a few lines as there is no need to compute the md5 hash when we're
returning a data URI. Previously md5_file was called twice during module builds
(once for the fallback url, and another time when producing the embedded data uri).
Applied to logo in SkinModule as example. To be applied elsewhere as needed.
Without it, fallback is current behaviour (no cache invalidation).
Bug: T99096
Change-Id: I7f38bfc1bea5c241bc4f8ec4f4b640fd65f2c04f
If an extension specifies a require section in its composer.json, which contains
some dependencies, we can assume that this extension needs to load the composer
autoloader.
The maintenance script convertExtensionToRegistration.php will now check if the
extension has such a composer.json and, if so, adds the load_composer_autoloader
property in extension.json, which will try to load the autoloader of composer
is it is present.
Also add a check for existence of a require section in ComposerJson library
(ComposerJson::getRequiredDependencies()).
Bug: T119766
Change-Id: Icdbc37abc44e642afee2aab4c0e9298d3471124d
* This can happen due to incr/add races. Use incrWithInit()
instead to handle such cases.
* Also made BagOStuff:incrWithInit() return the new value like incr().
Change-Id: I0e3b02a4cff7c20544a9db2eaabd3f61e5a470b1
CURL has a "feature" where passing array( 'foo' => '@bar' )
in CURLOPT_POSTFIELDS results in the contents of the file named "bar"
being POSTed. This makes it impossible to POST the literal string "@bar",
because array( 'foo' => '%40bar' ) gets double-encoded to foo=%2540bar.
Disable this "feature" by setting CURLOPT_SAFE_UPLOAD to true,
if available. According to the PHP manual, this option became
available in 5.5 and started defaulting to true in 5.6.
However, we support versions as low as 5.3, and this option
doesn't exist at all in 5.6.99-hhvm, which we run in production.
For versions where this option is not available (pre-5.5 versions
and HHVM), serialize POSTFIELDS arrays to strings. This works
around the issue because the '@' "feature" only works
for arrays, not strings, as of PHP 5.2. (We don't support pre-5.2
versions, and I've verified 5.6.99-hhvm behaves this way as well.)
Bug: T118032
Signed-off-by: Chad Horohoe <chadh@wikimedia.org>
Change-Id: I3f996e2eb87c7bd3b94ca9d3cc14a3e12f34f241
APCu's object serialization causes memory corruption when apc.serializer is set
to 'default' (see https://github.com/krakjoe/apcu/issues/38). We can work
around this bug by falling back to the pre-I4b2cf17155 behavior of
APCBagOStuff, which is not to trust apc_store() with anything other than
strings and integers, and instead serialize everything in user-space PHP code.
Bug: T120267
Change-Id: If34a1d959e2700792514b372af52919940222d83
Currently Timing::measure() does not check that the requested start and end
marks exist, causing it to return bogus values without any indication that
something has gone wrong. Fix this by logging and error and returning false
in case either the start or end markers do not exist. To make it possible to
log, make Timing implement the LoggerAware interface.
Change-Id: I75af5273e9a8a52b31d0af1de206b0d8a4c82fbc
MessageBlobStore class:
* Make logger aware.
* Log an error if json encoding fails.
* Stop using the DB table. WANObjectCache supports everything we need:
- Batch retrieval.
- Invalidate keys with wildcard selects or cascading check keys.
* Update tests slightly since the actual update now happens on-demand as
part of get() instead of within updateMessage().
ResourceLoader class:
* Remove all interaction with the msg_resource table. Remove db table later.
* Refactor code to use a hash of the blob instead of a timestamp.
Timestamps are unreliable and roll over too frequently for message blob store
because there is no authoritative source. The timestamps were inferred based on
when a change is observed. Message overrides from the local wiki have an
explicit update event when the page is edited. All other messages, such as
from MediaWiki core and extensions using LocalisationCache, have a single
timestamp for all messages which rolls over every time the cache is rebuilt.
A hash is deterministic, and won't cause needless invalidation (T102578).
* Remove redundant pre-fetching in makeModuleResponse().
This is already done by preloadModuleInfo() in respond().
* Don't bother storing and retreiving empty "{}" objects.
Instead, detect whether a module's message list is empty at runtime.
ResourceLoaderModule class:
* Make logger aware.
* Log if a module's message blob was not preloaded.
cleanupRemovedModules:
* Now that blobs have a TTL, there's no need to prune old entries.
Bug: T113092
Bug: T92357
Change-Id: Id8c26f41a82597e34013f95294cdc3971a4f52ae
It was already used everywhere as non-static via $this.
This is needed in order to allow MessageBlobStore unit tests
to disable the holdoff via a mock (mocks can't override static methods).
Change-Id: I3aad5b6e780addf1b6ce9de56c81b91f5ab358b2
If a set() happened around the exact same time as check key was
initialized via get(), the curTTL in future get() calls could
sometimes be 0 instead of a negative value, even though hold-off
should apply.
Change-Id: Ide1fd65112aff425a4798e2ec406d71f2a8e84a7
Move the holdoff period into the purge value instead of deciding
it at runtime. This opens the way for touchCheckKey() to support
a custom $holdoff parameter, which will allow callers to invalidate
keys without a holdoff period. Similar to what we already support
in delete().
Right now the holdoff period is decided at run time.
Change-Id: Id10c036272e92ae4429effc823b75e08fb11a48b
Doc block has also been changed to show that
this implementation does not follow the defined
interface.
It also says why...
Bug: T116031
Change-Id: If8cca355ae77b751486b40e56a6b3a51b1d7fe0b
There is no strong use case for associative "decoration"
of constructor arguments (the documentation benefits
mentioned in I43aa085 are outweighed by the confusion
caused by not failing loudly when someone passes an
associative argument by accident, e.g. by omitting an
array nesting level), so disallow them but make sure
they fail nicely, not with an invalid offset error.
Change-Id: I09e4af85ded6a1497b0db0265d2ee6707f91f5e3
Arguments are sometimes given as array( 'foo' => 1, 'bar' => 2 )
which makes the configuration self-documenting.
Change-Id: I43aa085090f1014ba841641867ebf9559d16e76d
* Code blocks must be indented by 4 spaces (so 5 in total from the star).
* Code blocks are only supported in descriptions, and for some reason the first
line is separate from the description, so there must be an additional line of
text between the summary and the code block since it can't be the first thing
in a description.
* Fix regex confusion with comment syntax due to "/*\/".
To verify these findings, use:
$ php maintenance/mwdocgen.php --file includes/libs/StringUtils.php
And check the output warnings and generated html pages.
Change-Id: I259b87c0bf3b48a95df0108eb6ee799c25b2853f
To allow batch queries for multiple keys that themselves have different check
keys. Previously check keys always applied to all keys being retrieved.
Change-Id: I9e5ba198d79020ce05a802a510762e29fcfb2f1b
Follows-up f36b73e96c, which moved these classes to libs/objectcache.
* Fix wrong @throws in MemcachedPeclBagOStuff.
* Fix wrong @returns in MemcachedBagOStuff::getClient().
* Rename MWMemcached to MemcachedClient.
* Remove mention of 'memcached.php', which doesn't exist anymore.
Change-Id: I34dbc859be4778cea489fd2344f233f30452605c
* Add `Timing`, an interface which mimics the W3C User Timing API.
It provides a canonical way to store and retrieve markers (timestamps)
and measures (timestamps + duration).
* As the initial use-case, use it to record 'requestShutdown'.
Change-Id: I36b29162ffcc091406df025463b0e2797e52f19a
* Match behaviour of MapCacheLRU and ProcessCacheLRU.
* Add missing unit tests for TTL and maxCacheKeys eviction behaviour.
Change-Id: I559eae1cd336274b21728e86775cfbad7e2f2c6d
Especially useful when interacting with getWithSetCallback of
HashBagOStuff where a $exptime is required before the callback.
In set() it is typically left out and thus not as counter-intuitive.
Change-Id: Ic2f7adda3c00cefe701d77bea91a7e8e77ef1439
