- mostly auto fixes
- some too long lines fixed
- ignore amp space in one case passing by reference
Change-Id: I6472f83bc3cbf4bd629d83050cc3319b19ec465c
Having such comments is worse than not having them. They add zero
information. But you must read the text to understand there is
nothing you don't already know from the class and the method name.
This is similar to I994d11e. Even more trivial, because this here is
about comments that don't say anything but "constructor".
Change-Id: I474dcdb5997bea3aafd11c0760ee072dfaff124c
Undo traces of a practice we carried over from past projects and
existing examples that is neither universal nor actively encouraged in
the MediaWiki codebase.
Bug: T139301
Change-Id: I5c9c89b72a45a44aa4264a5e57b003c1a86cdf6e
Co-Authored-By: Brad Jorsch <bjorsch@wikimedia.org>
The used phpcs has a bug, so the version 0.9.0 could not be enforced at the moment.
Will be fixed in next version, see T167168
Changed:
- Remove duplicate newline at end of file
- Add space between function and ( for closures
- and -> &&, or -> ||
Change-Id: I4172fb08861729bccd55aecbd07e029e2638d311
It's unreasonable to expect newbies to know that "bug 12345" means "Task T14345"
except where it doesn't, so let's just standardise on the real numbers.
Change-Id: I6f59febaf8fc96e80f8cfc11f4356283f461142a
xhprof does not support PHP7, and it doesn't seem like upstream will be
working on that any time soon.
Tideways is a profiler that is basically a drop-in replacement for
xhprof with functions renamed. So Xhprof::enable() and Xhprof::disable()
will now try to use tideways if that is installed and xhprof is not.
Bug: T152186
Change-Id: I0d7d2de56ac638ca2851c662f527049bd620c0e9
Avoids "Argument 1 passed to Profiler::scopedProfileOut() must
be an instance of ScopedCallback, SectionProfileCallback given"
Change-Id: I92713de71df9722d8a5d7e5cd04460aff71cf096
The ScopedCallback class was moved into a separate library. This updates
all callers to use the namespaced version, and provides a
backwards-compatibility class wrapper under the old name.
Bug: T146258
Change-Id: I2dd0a66fe2f510f26bdfef6b0a975c1beb3fd93c
Depends-On: Iea0c40bdd7776372ccf72db8f088a2abaa4d3721
The motivation for this patch came from trying to use xhprof to profile the
unit tests. I was able to profile specific test suites, but if I tried to
profile a complete PHPUnit run, I ended up with empty profiling data. My
initial suspicion was that this was due to some Xhprof buffer getting
exhausted. The actual reason ended up being much simpler: the XhprofTest suite
indirectly called xhprof_enable() / xhprof_disable(), which stopped xhprof and
cleared out the data, so that when I was calling xhprof_disable() at the end of
the run, there was no profiling data to return, because xhprof was not running.
For the most part the XhprofTest was already doing the right thing by trying to
avoid having side-effects or relying on xhprof. Wherever possible, test fixture
profiling data was used in lieu of actually running xhprof. But this was not
totally successful because the Xhprof class coupled the collection of data to
the processing of data. Xhprof::__construct() called xhprof_enable(), so there
was no real way around that.
I think that the right way to fix that is to decouple profiling from profiling
data analysis. Thus I renamed 'Xhprof' to 'XhprofData', and modified the class
so that it expects to be fed profiling data rather than going out and
collecting it on its own. As a result, it is now possible to profile a full
phpunit run with xhprof, and the work that went into writing fixtures for the
Xhprof unit tests pays off: the class and the tests no longer have a hard
dependency on the xhprof extension, and the tests do not have to be skipped
when it is not installed. And the tests are really testing the system under
test, rather than the xhprof extension.
Finally, I added a new Xhprof class, which really is just an extremely thin
wrapper around xhprof_enable() / xhprof_disable(). The only extra functionality
it provides is the ability to check whether xhprof is running, via
Xhprof::isEnabled(). Calling Xhprof::enable() when it is already enabled will
cause an exception to be thrown. This should help us avoid running into
situations where two components contend for control of the profiler without
realizing it. A unit test tests this behavior.
The only part of this change that is not covered by tests is the change to
ProfilerXhprof. I tested it manually and it works.
Change-Id: Ica96beded68f04083abaf48ef1ae8c726eb60fa6
All unused:
* $wgProfileCallTree
* $wgProfilePerHost
* $wgUDPProfilerHost
* $wgUDPProfilerPort
* $wgUDPProfilerFormatString
* $wgStatsMethod
* $wgAggregateStatsID
* $wgStatsFormatString
The only one still used was $wgProfilePerHost.
Also move release notes for ade945b97a from HISTORY to RELEASE-NOTES (Oops!)
Change-Id: Ia0280d23674ace1ac33aa561b06ee73cab865b29
This uses a non-standard output and requires a custom
collector that wmf does not maintain nor use anymore.
Change-Id: I41a68f7061465417fbdc5ca41f8eb6e1f99f1111
RUSAGE_THREAD is not POSIX; it's Linux-specific. HHVM now compiles on OS X,
and calling getrusage( 2 ) on OS X explodes with:
exception : f_getrusage is not supported: RUSAGE_THREAD is not defined on
this sytem [sic]
HHVM is being ported to Windows, too, where RUSAGE_THREAD is likewise missing.
So check that PHP_OS === 'Linux' before using it.
Change-Id: I6eac7d2bb4edfa1e697ae72f89299d0d5ac135f4
I'm not sure why I stuck `normalizeMetricKey' in ProfilerOutputStats, because
the transformation it applies are suitable for converting any arbitrary string
into a StatsD-safe metric key. This patch moves the method to
BufferingStatsdDataFactory, which ensures it applies to all metrics logged
within MediaWiki, and not just the Profiler.
Supercedes If0237cdd0d.
Change-Id: I496ed748000d28f5399fee6e3cc271a1f68bd058
See https://blog.wikimedia.org/2015/06/12/securing-wikimedia-sites-with-https/
Ignoring SVN stuff because svn.wikimedia.org's SSL certificate has expired, and
ignoring wgSharedUploadPath because T25122 wants to get rid of it.
Also a couple of BZ -> phab changes
Change-Id: I02fd23ac6c30a4a4c718e57d4dedbc693dd653c0
* Added doPreOutputCommit() and doPostOutputShutdown(),
which most entry points just using the later
* Also fixed problem where text profiling did not show up
* Avoid calling triggerJobs() in the file streaming
entry points
Bug: T100127
Bug: T100085
Change-Id: Ibc7e768fd483389a01847f08cdeba4058c853d3f
* Apply the blacklist / whitelist to profiled sections, not just function names.
* Allow shell-style wildcard patterns in blacklist / whitelist.
* Prefix all profiled section names with 'section.', to distinguish them from
functions.
Note that shell-style wildcard patterns are not supported by xhprof natively,
but it won't barf on them either, nor will they match against actual function
names (since shell wildcard characters are not valid for PHP function names),
and the filtering will still be enforced in ProfilerXhprof.
This has the side-effect of working around https://github.com/facebook/hhvm/issues/4385
Bug: T99829
Change-Id: I8354ed922fa7b42857eda03be8f62b89ac78d0d6
If one wants to nest all metrics emitted by the profiler under a metric
namespace, one can now set the 'prefix' param.
Task: T66301
Change-Id: I6c52f20e39017f4c818ca6623bb7f48683fc8abc
When we convert a profiler entry to a statsd metric key, we replace
nonalphabetic characters with slashes. When part of the key is a file path with
a leading or trailing slash, this can result in keys that look like:
'MediaWiki._srv_mediawiki_foo_.bar', which is a bit ugly. This patch trims
leading and trailing underscores from each dot-separated key segment.
Change-Id: I71172b956f5a6120a75981a15dedf372d3893b88
* This makes it possible to see what is being done by matching
the trx IDs of the (normally) adjacent log entries
Change-Id: I515de99acfe159856930d8e259c00eb744f43213
This will allow us to avoid needlessly investigating master requests that
only happen on POST (which is already what we want for multi-DC).
Bug: T92357
Change-Id: Ia7437d00f5b89a8e318d85659d60e2f9f9f26149
Instead of maintaining a mapping of short names to class names ('db' =>
'ProfilerOutputDb', etc.), let us adopt the convention of using the full
class name to indicate the output type. We can maintain backward-compatibility
by using simple string manipulation to transform short names to the full class
names.
Change-Id: I976e0da2873d88b9892fb41823cfe3af0a2d3974
* 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
