There are imho too many ObjectCache methods, $wg*Cache* variables,
CACHE_ types, and wfGet*Cache() functions to know which ones
should actually be used publicly. This should make it easier to
know which ones should be used.
The difference between them was also harder to understand with the
main documentation sections about each of these living in
conceptually different places:
- classes (WANObjectCache)
- methods (ObjectCache::getMainStashInstance)
- variables (DefaultSettings: wgMainCacheType, wgMainWANCache, wgMainStash)
Also:
* Correct wfGetCache() documentation.
* Add the missing keywords for 'public' visibility.
Change-Id: I3b05fdb8b7888bf7e3f05bdca36538f3484556b8
* Potentially long running POST requests often use multiple transactions,
talk to multiple services, or defer updates. Try to make sure they have
a chance to complete all of the work. WMF already sets ignore_user_abort()
across the board in config, but this applies it to key spots for all
installs, in addition to bumping the time limit.
* Eventually this can lower the need for high overall time limits.
Bug: T102890
Change-Id: I893ddd773064dcd63b5b24c84c6391974f4b5aee
* This should trigger more reliably than the shutdown function
callback, which is really only there for sanity to make sure
session close/write happens.
Change-Id: I9a1aa76de121ba8de33b3fa850bd223929fae404
As of 155d555b83, we now redirect variations on hex escapes
into their canonical form. This was causing '~' to be redirected
to %7E. However google chrome seems to canonicalize %7E back
into ~, causing a redirect loop.
RFC 3986 says ~ is unreserved, so not hex encoding it should be
fine. To quote: "For consistency, percent-encoded octets in the
ranges of...tilde (%7E) should not be created by URI producers"
Bug: T105265
Change-Id: I01556eee496e2fb540de8ff09c082c1fedddb5f7
strtr() is marginally faster as it runs through the string only
once. A better fit for one-for-one character translation.
The strtr() function also supports an associative array as second
parameter for entire string replacements. This, too, has the same
performance and predictable behaviour (starts with the longest key).
Whereas str_replace is for more aggressive needs where you want
multiple passes until there are no further matches.
The associative array form is arguably also easier to understand
and harder to mess up since the needle/replacement pairs are
explicitly connected instead of two separate arrays.
Also:
* Use getFormattedNsText instead of strtr( getNsText, .. ) which
reduces duplication of this fact through a more semantic intent.
Change-Id: Ie23e4210a5b6908dd79eebc8a2b931d12fe31af6
Some code paths abuse wfForeignMemcKey() for this purpose. This is semantically
incorrect and seems fragile. Also the empty second argument (for db-prefix) is
either misused or akwardly set to the empty string.
It also creates a namespace conflict between arbitrary application keys (from
the software) and any database names (from users). This commit reduces the
conflict surface down to a single key (namely, "global").
Also added unit tests to assert the implicit restrictions and
assumptions between these cache key functions are valid.
Change-Id: Ia0953b51005fe3de3b881dd1bd64c9d3c85e8c66
wfSuppressWarnings() and wfRestoreWarnings() were split out into a
separate library. All usages in core were replaced with the new
functions, and the wf* global functions are marked as deprecated.
Additionally, some uses of @ were replaced due to composer's autoloader
being loaded even earlier.
Ie1234f8c12693408de9b94bf6f84480a90bd4f8e adds the library to
mediawiki/vendor.
Bug: T100923
Change-Id: I5c35079a0a656180852be0ae6b1262d40f6534c4
Also:
* Update wfHttpError() to use uppercase DOCTYPE, to match other code
such as Html.php, wfThumbError(), HttpError.php, etc.
Change-Id: I4027e7fe1a138b03f78797b6d1bfe7bd1064d360
The warning occurs when gmp_init() tries to convert the empty string
that came from ltrim(). This is causing tests to fail under HHVM 3.6.1.
Follows-up 5957856c46.
Bug: T98882
Change-Id: Ica86c91d7897db979e09d7cfc82fb3a20b95d4ce
Also sets default paths immediately (not in Setup.php) so
they are available before extensions register.
Bug: T98319
Change-Id: I41a8aec7a3e9c576ec7344abf51f8106248ade4b
A user just logged on to the #mediawiki channel to ask for help because his
webhost disallowed php_uname(), causing wfIsWindows() to crash. Google
autocompletes 'php_uname()' to 'php_uname() has been disabled for security
reasons', so it is probably not uncommon. Consulting the PHP_OS constant
instead side-steps the problem nicely.
Change-Id: I8d63826db4fc5d142eac53717d4f9fbbf9928de9
* Previously if it was called before a DB_SLAVE connection, it would not
be set. This is fixed now.
* Also set mLaggedSlaveMode in LoadBalancer as appropriate. Before, it
was only set in the "too lagged for ChronologyProtector" case.
Change-Id: Ic4dc555cf762653f157df1795f53f3577c1e587a
Move the MWLogger PSR-3 logging related classes into the
MediaWiki\Logger namespace. Create shim classes to ease migration of
existing MWLoggerFactory usage to the namespaced classes.
Bug: T93406
Change-Id: I359cc81fbd2dcf8937742311dcc7d3dee08747b0
* Associate Profiler objects with a request context by adding a $context
property with a getter and a setter.
* Introduce ProfilerOutputStats, which writes profiling data to the stats
buffer associated with the current request context.
* Make it the Profiler class's responsibility to enforce $wgProfilerLimit.
* Deprecate $wgProfilerLimit in favor of the (more aptly named, IMO)
$wgProfiler['threshold'] config setting.
* Tidy up Profiler instance creation code in Profiler::instance().
* Add Profiler::getOutputs, which returns an array of ProfilerOutput instances
which are configured for the current profiler and whose canUse() method
returns true.
* Make ProfilerStub not log by creating a stub ProfilerStub::logData() method
which does not call the parent. Previously the parent class checked if $this
was an instance of ProfilerStub and returned early if so.
Task: T90623
Task: T85641
Change-Id: Icf644ad3435c1f30d0a49957a97b481808a3153d
* Deprecate $wgRequestTime in favor of $_SERVER['REQUEST_TIME_FLOAT'], which is
more accurate. Because $_SERVER['REQUEST_TIME_FLOAT'] is only set for PHP
5.4+, set it to microtime( true ) in WebStart.php for back-compatibility.
* Add a 'requestTime' property to WebRequest objects, set to
$_SERVER['REQUEST_TIME_FLOAT'] for WebRequest or the instance creation time
for FauxRequest instances.
* Use that to provide WebRequest::getElapsedTime(), which gets the time since
the request was initiated.
* In wfLogProfilingData(), get the user and request objects from the context
object rather than from global scope.
Opportunistic clean-up: move the magic quotes check to WebStart.php and make
the error message more helpful.
Change-Id: I7e07e22eaf16b5141b80ad9f843285c542a127b7
Right now wfLoadExtension() and related functions explicitly load
extensions immediately, bypassing the queue. This was done to be
extremely backwards-compatible with the old require_once style of
loading which does the same.
However, for a future configuration database to work, we need to be able
to reliably load extensions after configuration (LocalSettings.php) is
loaded, which is currently at the top of Setup.php. Rather than doing
this later, we should do this now to make sure the registration system
will be able to handle it.
In Wikimedia production, excentions are currently being loaded with
direct calls to:
ExtensionRegistry::getInstance()->queue(...);
so we know that this should work, but that is not a nice API for sysadmins
and developers to be entering into LocalSettings.php.
If for some reason an extension really needs to be loaded immediately,
they can still call:
ExtensionRegistry::getInstance()->loadFromQueue();
But that should be the exception, not the norm.
Change-Id: I72672e5c9541ede02d09f548c39ef6c8df0ec78a
This function is backwards:
* The stated purpose is to generate a list of differences for callers to parse,
however we already have DiffEngine that returns a list of differences that you
don't need to parse manually shooting your feet off in process.
* With improvements in PHP performance in the last 10 years, shelling out instead of using
a pure-PHP diff has a doubtful performance benefit.
* Even in configurations where it's faster on predominant diffs (which have to be huge),
shelling out has its own security implications.
This function is not used in core. The only extensons using it are Echo (which just parses
its output so would be better off with DiffEngine) and AbuseFilter, let's not add more users
by tolerating this function.
Also, add error checking while I'm at it.
Change-Id: Ia67debce39de8252312fd887ebfbe6fb89f9edc9
Refactor out 'infinity' vartiant values which used in blocking and
protecting actions. This patchset adds GlobalFunction wfIsInfinity.
Bug: T68646
Change-Id: I60cc55a5bbd43c72916a1c2ea3807457d4e33765
This patch adds a metric data service object to the IContextSource interface,
with full support for StatsD meters, gauges, counters and timing metrics, via
the liuggio/statsd-php-client, which this patch also introduces.
Usage example:
$stats = $context->getStats();
$stats->increment( 'resourceloader.cache.hits' );
$stats->timing( 'resourceloader.cache.rtt', $rtt );
The metrics are flushed to a StatsD server, which may be specified via the
'StatsdServer' configuration key. If no such configuration key exists, the
metrics are discarded.
The StatsD client supplants MediaWiki's StatCounter class. wfIncrStats()
will continue to work, but it will delegate to the StatsD data object.
Change-Id: Ie10db1c154d225971398e189737de7c560bf0f90
wfEscapeShellArg() can handle multiple params, escaping each.
This patch changes wfShellExec() to call wfEscapeShellArg() directly
instead of doing the gluing itself.
This patch also extends wfEscapeShellArg() to accept an array parameter
optionally instead of as separate args, which is often useful.
Added also unit test cases for single, multiple args, and single array args.
Change-Id: I7a0761cc2ba98c210a9eacadd12da407d933e42a
Instead of instantiating this on every single request. Removes
wfGetLangConverterCacheStorage() and $wgLangConvMemc which were
otherwise unused.
Change-Id: Ic500944a92c2a94bc649e1b492c33714d81dca00
If $wgResponsiveImages is on, include the 1.5x and 2x multipliers
in the wfThumbIsStandard() standard image size comparisons.
Change-Id: I79d866387730cdd4d7c7a976e691192b2335b7ce
Time wounds all heels. During the code review for the PSR-3 logging
introduction, several people asked me why we needed a wrapper
for Psr\Log\LoggerInterface if the point was to use the standard. At the
time I was convinced that it would be better to introduce the dependency
via a wrapper class so that we could use the wrapper to patch over any
deficiencies that we might find in the PSR-3 API. After going on to work
on a project to disentangle other MediaWiki components from internal
project dependencies I have suddenly and clearly seen the error of my
ways.
We still need a logger factory as PSR-3 does not specify
a standard mechanism for creating Psr\Log\LoggerInterface instances. My
solution is to convert MWLogger into MWLoggerFactory to retain a static
factory interface for creating PSR-3 loggers but remove the MWLogger
wrapper class itself in favor of direct exposure of
Psr\Log\LoggerInterface to the MediaWiki consumer classes.
Change-Id: Ie47467657dcf341991ada00827dca5e8eff95438
For example, if the database cannot be connected to, the language is
something other than 'en', and the l10n cache is LCStoreDB, then the
call to Language::factory() in Setup.php line 604 will do it when it's
trying to get the fallback list.
Bug: T86398
Change-Id: Ia0e9ea440dcc4c8ab7bab1f33802aeff487eb3f0
