No direct references found in core or extensions outside of the classes
themselves. None of these are subclassed, so set all of them to private.
Change-Id: Id599c0830c1e6d3679c7ae55147f4444f8252987
Mark all private, except the one that's being used in subclass. Checked
core and extension for direct uses. None found.
Change-Id: If73d65ca094ff56deb19d3b6f3ef99892654726d
iPart of program to remove underscores from class names. Checked core and
600+ extensions for occurrences. All uses are in core, and are updated in
this change.
Change-Id: I432dc249d22053728013ae7d0d56c3c398021c5e
Part of program to remove underscores from class names. Checked core and
600+ extensions for occurrences. All uses are in core, and are updated in
this change.
Change-Id: Ib157c3cf209677718ff184552cf872b1d3f4e969
* In partition(), avoid doing the JOIN since it is not needed and prevents
index-only queries.
* Made numLinks() wrap partition when no $max is specified.
* Also fixed some docs
Change-Id: I05a83f71efb4c6e99e40883d7fa53da59184b13c
The grouping makes at least as much sense as job/, and certainly makes
more sense than cache/. With directories named after base classes, it is
fairly easy to tell what should go where. The grouping of
DeferredUpdates, DataUpdate and CallableUpdate would surely be
uncontroversial.
The move of SearchUpdate out of search/ demonstrates the conflict between
arrangement by module versus arrangement by type, which is the most
difficult design question here. I think arrangement by type is more
consistent with e.g. the arrangement of the core root, i.e. tests/,
resources/, maintenance/, etc. where a given feature will have its files
split up into a mostly type-based hierarchy.
I also tidied up AutoLoader.php by moving includes/content to the correct
location, sorted alphabetically by subdirectory.
Verified with AutoLoaderTest.
Change-Id: Ib369411d0caca38e72978084aa57348f1b892ed0
The existing "linkprefix" message is unlikely to be accurately
customized by message translators (as shown by the fact that, of the 10
distinct customizations prior to Iaa7eaa44 (which made them even more
complicated), 3 were broken or entirely ineffective, 1 was half
ineffective, and 2 more seem to have included the Latin-1 Supplement by
accident) or by local wiki admins. So, like linktrail before it, let's
move it out of the system messages and into a separate language
variable.
At the same time, let's make it a simple character set (like
$wgLegalTitleChars) rather than a complicated regular expression. The
complicated regex now lives in the parser.
This also adjusts the output of the API's action=query&meta=siteinfo and
adds an accessor parallel to the linkTrail accessor to Language.
Note the following changes that are not simply extracting the existing
charset from the linkprefix message for $linkPrefixCharset:
* The En message matched all non-ASCII UTF-8 characters by matching the
component bytes (\\x80-\\xff). The new character set is equivalent.
* Various languages were identical to En and so have no $linkPrefixCharset
set. These are: Ary Az Ce Ga Id Ka Kiu Km Ltg Mk Ms Ne Nn Ro Roa_tara Sc Si
Sr_ec Sr_el Tl Tt_cyrl Tt_latn Ug_arab War
* Cu, Uk, and Udm are changed to match any number of „ or « in the prefix.
* Cv tried to include "«" that was redundant to the range \\x80-\\xff
(see En comment). This was removed.
* Diq was entirely bogus, and so was removed.
* Gu included many additional UTF-8 characters that are redundant to the
range \\x80-\\xff (see En comment). These were removed, and the
resulting character set is equivalent to En.
* Mt has been broken since it was introduced in r37242. The charset used is
equivalent to the broken regex.
Bug: 56031
Change-Id: I3369851b33113fc118a1bace38f3ac310cdd9725
It's elitist mathematical jargon. In all cases dealt with here, it adds
no additional meaning compared to "if", beyond what was already obvious
from context. Thus, its only purpose is to smugly demonstrate that the
author attended their second-year mathematics classes, at the expense of
causing confusion for everyone who doesn't have such a background.
If you really think you need to convey extra information beyond what
"if" gives you, the English language contains plenty of devices for doing
so, without resorting to neologisms.
Change-Id: Iae21095d02ec2935c10e94f532235c2671c115b1
Also removed some unnecessary ones. I think I've caught them all.
The spaceless version already appears in core ~300 times (after
accounting for false positives when grepping). Some consistency would
be nice.
Change-Id: I607655b5f4366e66dc78730d5fd2f57ed8776cae
This patch does two things:
A) rename $wgHTCPMulticastRouting to $wgHTCPRouting since you can have
MediaWiki send purge packets over unicast.
B) add support for multi hosts purge in the few cases one want to split
the multicast groups per cache role but still want to purge more than
one cache group.
A) rename
The rename adds deprecation notices in the comments and adds a back
compatibility layer in the case someone is already using this feature.
Given Wikimedia Foundation is not even using it, it is very unlikely,
but yet: better safe than sorry.
My logic is flawed sometime, so that needs a bit of review :) There is
two levels of deprecations to watch for:
- pre mw1.20 which useds $wgHTCPMulticastAddress and
$wgHTCPMulticastPort
- pre mw1.22 (aka before this patch) that used $wgHTCPMulticastRouting
The resulting configuration should be properly loaded in $wgHTCPRouting.
B) multi hosts
The HTCP routing let you split purges to different hosts according to
the URL. In some case, we want to be able to purge an URL from more
than one multicast group. A Wikimedia example would be to send text
related purges to the text and mobiles caches while upload purges are
sent to the upload caches.
An abstracted example would be:
$wgHTCPRouting = array(
// upload URLs to upload caches
'/(upload|thumbs)/' => array(
'host' => '<ip for upload caches>'
),
// Everything else to text & mobile
'' => array(
array( 'host' => '<ip for text caches>' ),
array( 'host' => '<ip for mobile caches>' ),
),
);
Change-Id: Ie87f6b81007f47f9cb439371743f3ad9a4b0c774
I tend to dislike code style such as:
if( $ok ) {
// lot of code
} else {
return false;
}
The SquidUpdate::HTCPPurge() used that pattern to handle a
socket_create() issue. I have simply moved the block above and inverted
the condition. Ie:
if( ! $ok ) {
return false;
}
// lot of code
That saves a level of indentation and makes the code a bit easier to
follow IMHO.
Change-Id: Ic3c6e22bbb637352fef740a0c1663a4b6f5a2b6a
When a language code is unloaded, it also needs to clear the
shallowFallbacks entry. Otherwise the next time that language is used it
will incorrectly think that the other setup done by initShallowFallbacks
is still set up, leading to warnings.
Change-Id: Idf6a78f56a3bb864cc921427ac82972594610047
To debug purging issues in beta, it would be useful to have all squid
related messages grouped at the same place. wfDebug() is logged on beta
but it has a good amount of spam which is not that useful. Regrouping
all squid operations under the debug log group 'squid' will make it
easier to investigate.
Change-Id: I11112585f38307ae56ec3e05e7a8d6d9dab31bfa
Using the following command line, I have found doc comments mentioning
"1.21" when they should mention "1.22" instead, which I have fixed
manually:
git diff REL1_21 | grep --color=always -C 10 -iE \
'^\+.*(since|deprecated).*1\.21(\D|$)' | aha > oldver.html
I also moved the release notes for I1987190f ("Combine JavaScript and
JSON encoding logic") from RELEASE-NOTES-1.21 to RELEASE-NOTES-1.22
because I had reverted the commit on REL1_21 only (see Id3b88102 and
bug 47431 for the rationale).
Change-Id: I11b917a371e07267dfa98b8449776d0c1cb29b15
Follows-Up: I25cf5a94f6e47f85a9d0b80cc1c9c9f957288478
Follows-Up: I3d72e4105f6244b0695116940e62a2ddef66eb66
Follows-Up: I3faa9c3e8107c6e46cdf21f8c18adda1f42890d7
Follows-Up: I6aab19c8d68bf47beddad42632b0360a7b12f251
Follows-Up: I86368821fc2cd0729df5342b8572eb470c0f77a0
Follows-Up: Id3b88102e768318e3605a19e9952121091a40915
Follows-Up: Ie667088010e24eb6cb569f9e8e8e2553005223eb
Functions expect the message cache to return a string if a message
exists, even empty, and false if it does not exist. This adds casting
to the substr() function, which would return false for existing
messages that were just blank.
Bug: 14176
Change-Id: Id91914a3701fe53f1e2e894824512489392c628b
This patch adds the ability to destroy the LinkCache singleton or to
replace the instance with a different object. Note the $instance static
variable is now at the class level instead of at the method level.
Change-Id: Ib110ad061868834a52a6bac651976b3ffab4a6ce
The core function behind wfMessage() (MessageCache->get()) did not
apply the language fallback chain to on-wiki messages.
This patch has changed the behavior to iterate over all possible
languages, first checking on-wiki and then checking the CDB cache,
until it finds the message. Note that fallback languages never
take precedence over the actual requested language.
This patch was taken from the following changes and then
adjusted to fix issues that caused bug 46579.
* Change-Id: Iaaf6ccebd8c40c9602748c58c3a5c73c29e7aa4d
- Author: Matthew Walker <mwalker@wikimedia.org>
- (cherry picked from commit d434bfcf3b)
* Change-Id: Ib607a446d3499a3c042dce408db5cbaf12fa9e3d
- Author: Mormegil <mormegil@centrum.cz>
- (cherry picked from commit 1b8cb8dc31)
Bug: 1495
Bug: 46579
Change-Id: I420457863eeb79824698d06abc7784032b267af2
[includes/cache/MessageCache.php]
- internal constants
- constructor
- a few missing @var
[includes/clientpool/RedisConnectionPool.php]
- group internal settings applying to the pool
- misc protected members updates
[includes/debug/Debug.php]
- missing parameter name in @var statements
Change-Id: I6ff0a68d659529d128f40f32b0fd1c1d39af952f
Added spaces after/before parenthesis
Removed unneeded parenthesis around some statements
Broke a long line
Change-Id: I7fbe129f7bbf524dd0598ece2a9708643f08453b
Squiz.WhiteSpace.LanguageConstructSpacing:
Language constructs must be followed by a single space;
expected "require_once expression" but found
"require_once(expression)"
It is a keyword (e.g. like `new`, `return` and `print`). As
such the parentheses don't make sense.
Per our code conventions, we use a space after keywords like
these. We appeared to have an unwritten exception for `require`
that doesn't make sense. About 60% of require/include usage
was missing the space and/or had superfluous parentheses.
It is as silly as print("foo") or return("foo"), it works
because keywords have no significance for whitespace between
it and the expression that follows, and since experessions can
be wrapped in parentheses for clarity (e.g. when doing string
concatenation or mathematical operations) the parenthesis
before and after basiclaly just ignored.
Change-Id: I2df2f80b8123714bea7e0771bf94b51ad5bb4b87
Disable of MessageBlobStore clear
Reset $wgAutopromote (should be moved to a config change!!)
Disable setting of wgStyleSheetPath
Throttle page_touched
Add apc htcp packet numbers to SquidUpdate
Disable set names binary/utf8
Commment out searchindex table indexes
Was c532e81d583d3d0439fe76eea4d105d675461b56
Original revision Change-Id I42c4f859e55eb198f6c6841e582b3552aad7b31f
https://gerrit.wikimedia.org/r/#/c/7606
Change-Id: I5ec8dd53188e9e4128f99ceaff38ebf9dcf570bb
I noticed on special:listfiles/username, UserCache was doing queries
like select user_whatever from user where user_id in ( '1', '1', ...)
with the same user id 50 times (one for each result returned in the
special page). That seemed a little insane, so put the list of
users to query through a array_unique. (Quite likely the db would
optimize that query to not literally look up the same user 50 times,
but nonetheless it seems better to filter the list before then)
Change-Id: I80c8a359d0f7a53b2420ebcda641e594dd3c56e9
* This will do the page_touched UPDATE in autocommit mode to avoid
locking a bunch of titles for some unknown amount of time that
depends on how big the transaction is.
Change-Id: I6ec16a73c11fbf19c24d2b7fe7d01f91b9dcf33a
And added/removed spaces around some other tokens,
like +, -, *, /, <, >, =, !
Fixed windows newline style
Change-Id: I0b9c8c408f3f6bfc0d685a074d7ec468fb848fc8
Remove some trailing whitespace, break long lines, make PHPDoc consistent
and update it in places.
Change-Id: I340b43a47e121a44ebd4aaa2a45dff12e945e645
* Use the stale message cache while the new one is being generated
* Revert I811755d4 (make message cache load failure fatal). This
escalated several very plausible temporary site issues from barely
noticeable to complete downtime -- for example, memcached being down
on a site with only one memcached server.
* Remove $wgLocalMessageCacheSerialized, it's always been pointless
* Clarify a couple of comments.
* Increased lock wait timeout to 30s
* Make lock() fail immediately on memcached connection refused
Tests done:
* With local cache enabled: normal cold refill; refill local from
global cache; use stale local cache during remote refill; use stale
global cache during remote refill; cold cache wait for remote refill;
saveToCaches() failure; memcached connection refused.
* With local cache disabled: saveToCaches() failure; cache disabled due
to "error" status key; memcached connection refused.
Setting a 1-day expiry in memcached, with a ~10s CPU cost to replace, is
not the best idea since it inevitably leads to a cache stampede. Dealing
with the stampede by waiting for a lock is not ideal, even if it were
implemented properly, since it's not necessary to deliver perfectly
fresh message cache data to all clients.
This is especially obvious when you note that barring bugs, expiry and
regeneration always gives you back the exact same data, because we have
incremental updates (MessageCache::replace()). Keeping all clients
waiting for 10s just to give them the data they have already is pretty
pointless.
So, continue to serve the site from the stale message cache while the
new one is being generated.
One caveat: if local caching enabled, when the message cache becomes
stale, a sudden spike in network bandwidth may result due to the full
array (also typically stale) being fetched from the shared cache.
Bug: 43516
Change-Id: Ia145fd90da33956d8aac127634606aaecfaa176b
Follow up Ib94a8c18c4e270a7bd46faa17eb27a22ff950f75
Moved some empty lines to match the places of other profile calls in the
file.
Change-Id: I6616b18027cdb9e70b43948aaff999a2e25b07e0
* Batch the queries to get the partitions.
* Improved caching of getNumLinks() with $max.
* Added a TTL to the ProcessCacheLRU use for sanity.
* Some small logic cleanups in partitionResult().
* Also cleaned up some code duplication.
Bug: 43452
Change-Id: I0b9710fe222b3d2cb4dc9ab2eeb0758873a8066c
This patch set reverts the following:
* Iaaf6ccebd8c40c9602748c58c3a5c73c29e7aa4d
* Ib607a446d3499a3c042dce408db5cbaf12fa9e3d
* Ic59fd20856eb0489d70f3469a56ebce0efb3db13 (partially)
Bug 46579 comment 17 describes a desired solution. In
If88923119179924a5ec091394ccab000ade16b3e we are working on a fix, but it is
taking longer than we anticipated. There was a deployment window planned
about now, but we didn't make it. It makes sense to revert for now, and commit
a proper solution somewhere next week.
Bug: 46579
Bug: 1495
Change-Id: Iac7ac4357dd80e8cdb238a6a207393f0712b3fe5
After Iaaf6cceb, MessageCache::get() goes through the fallback chain,
which is unwanted for conversion tables (for example, we don't want
zh-hans table for zh, where zh means "no conversion"). Since the only
needed feature is to fetch text from MediaWiki namespace (conversion
tables in PHP are stored somewhere else), it is now changed to use
MessageCache::getMsgFromNamespace() instead to avoid fallbacks.
Change-Id: I46e0be31c9c0fe0a6e4923fc1aff0fbbadbf1d67
* Removed spaces around array index
* Removed double spaces or added spaces to begin or end of function
calls, method signature, conditions or foreachs
* Added braces to one-line ifs
* Changed multi line conditions to one line conditions
* Realigned some arrays
Change-Id: Ia04d2a99d663b07101013c2d53b3b2e872fd9cc3
For example, find out which rule type should be applied for 5 items
in Arabic. The result would be 'few'.
This implementation should be non-disruptive and completely backwards
compatible (which is the main reason it isn't a lot simpler).
Change-Id: I3d72e4105f6244b0695116940e62a2ddef66eb66
Doxygen expects parameter types to come before the
parameter name in @param tags. Used a quick regex
to switch everything around where possible. This
only fixes cases where a primitve variable (or a
primitive followed by other types) is the variable
type. Other cases will need to be fixed manually.
Change-Id: Ic59fd20856eb0489d70f3469a56ebce0efb3db13
Added/removed spaces around logical/arithmetic operator
Reduced multiple empty lines to one empty line
Removed wrong tabs before comments at end of line
Removed too many spaces in assigments
Change-Id: I2bba4e72f9b5f88c53324d7b70e6042f1aad8f6b
The core function behind wfMessage() (MessageCache->get()) did not
apply the language fallback chain to on-wiki messages.
This patch has changed the behavior to iterate over all possible
on-wiki fallbacks (starting with the user's language) before
using the built-in language cache (CDB files). Previously we only looked
for the existence of an on-wiki message in the users's language.
Performance wise, using the 'ab' language ('ru', 'en' fallbacks)
MessageCache::get (Averaged over runs and calls)
New Code: ~8.5% TET (110us/call)
Old Code: ~6.5% TET ( 90us/call)
TET: Total Execution Time
Change-Id: Iaaf6ccebd8c40c9602748c58c3a5c73c29e7aa4d
Also make the IS NULL in upper case, looks better and sql reservered
words often written in upper case
Change-Id: I844ec37bbf3fb00d95a43dfd6f58db1e67724d53
Added/removed spaces after opening/before closing parentheses
Added a space after a comma
Removed unneeded parentheses in condition
Change-Id: I306091347ccaaf11dee0cdfda3019cb0c12be51b
Disable of MessageBlobStore clear
Reset $wgAutopromote (should be moved to a config change!!)
Disable setting of wgStyleSheetPath
Disable squid updates in HTMLCacaheUpdate invalidateTitles
Throttle page_touched
Add apc htcp packet numbers to SquidUpdate
Disable set names binary/utf8
Commment out searchindex table indexes
Was c532e81d583d3d0439fe76eea4d105d675461b56
Original revision Change-Id I42c4f859e55eb198f6c6841e582b3552aad7b31f
https://gerrit.wikimedia.org/r/#/c/7606
Change-Id: I5ec8dd53188e9e4128f99ceaff38ebf9dcf570bb
I suspect this is the primary cause we occasionally have
* Sidebar stuck in default values
* No gadgets at all
* Main page full of <int:custom-message> occurances
Change-Id: I811755d4c250800177c4ea9e509bf6e1d13443a0
* Added JobQueue::deduplicateRootJob() function which uses cache
records of the last time a "root" job was initiated for a task
in order to invalidate prior jobs for that task. For refreshLinks,
the "task" is basically "enqueueing the refresh jobs for title X".
* (bug 27914) Also added new Job::getDeduplicationFields() function
and made use of it with refreshLinks to exclude things like 'masterPos'
from duplicate job check comparisons for refreshLinks.
* (bug 27914) Always resolve refreshLinks2 jobs down to refreshLinks jobs.
For each affected pages, one of them will get their job popped
first, which will remove the duplicates for that page unless
one page is in a refreshLinks2 jobs and the other in refreshLinks.
* (bug 37731) Made LinksUpdate/HTMLCacheUpdate defer the large
backlinks query by doing it in an outer job.
* (bug 42065) HTMLCacheUpdate will no longer purge pages that were
already purged since the job was added.
Change-Id: I71b743e0a38e60a874ca856e80cb761bea06b689
* This way, we can actually get persistent cache hits for
the HTMLCacheUpdate queries rather than always hitting a DB.
* Also moved BacklinkCache under the /cache directory.
Change-Id: I0666ee575fb42675f1a7dd9cb52665f0a12a66a9
Callers often tend to end up calling getFileStat(), at least indirectly,
or in various successive function on the same path. This created RTTs
when the file didn't exist since negatives were not cached. This change
does the following:
* Cache definitive negatives (404s) in the process cache.
Nothing is cached on failure (like network problems).
* Ignore process cache entries after a brief time period
so long running scripts do not have overly stale entries.
Change-Id: I356bd9f48281e3c7e7a273778b2aca59c521a0c7
This was developed for translatewiki.net, which has but
millions of messages in MediaWiki namespace. To avoid
exploding the message cache, not all of them can be
loaded. Instead of logging what messages have been
requested, now it uses the fact that if the
message has not been customised in the
site language, there is most likely
no reason to load it in any other
language either.
Change-Id: I6dd81b3858acfd2b73332e46668d46015e99c748
* @licence -> @license
* Protects inline HTML by using double quotes, our inline comments uses
elements such as <h1> or <firstnameLastname@gmail.com>
* Commands in lowercase (@TODO -> @todo, @NOTE -> @note)
* removes @abstract and @static since doxygen detects them from PHP
code.
* various undocumented function parameters
* typos in parameters declarations
Change-Id: I62ad6fc124c355bf31acc780b9614a59cf79a421
Was accidentally broken in Ie38ae198, where the request URL was compared
against the final target title, not the title of the redirect. Also
tested squid purges of redirects, since that code presumably hasn't been
tested for a while, and committed my very advanced testing framework.
Change-Id: Ib2ffe8e109bcc4859e22df5470bddfdb09e67afa
Factor out duplicated code into the new ProcessCacheLRU. It is a fixed
size an per process cache which invalidate the least recently used cache
key.
Change-Id: Ib4475f21879ef6286ce2a28f248acd296fdbd45d
The content model is stored as a varbinary(32), the format
as varbinary(64).
If the standard model resp. format is used, null is written
to the database instead of the actual id, saving space.
Change-Id: I32659b49a9ad3cb8ecae9019562cff7de42b65f9
the queries for title and pageids are own queries created by the api,
which does not use the GenderCache, that can produce a query per row
Change-Id: I932f8d9d1cfa751dbb6f5237e2de325527d3ff53
Disable of MessageBlobStore clear
Reset $wgAutopromote (should be moved to a config change!!)
Disable setting of wgStyleSheetPath
Disable squid updates in HTMLCacaheUpdate invalidateTitles
Throttle page_touched
Add apc htcp packet numbers to SquidUpdate
Disable set names binary/utf8
Commment out searchindex table indexes
Was c532e81d583d3d0439fe76eea4d105d675461b56
Change-Id: I42c4f859e55eb198f6c6841e582b3552aad7b31f
LinkBatch can also give subpages to the GenderCache and therefor it is
easier to do it always in GenderCache, than in LinkBatch and Title
Add unit tests for GenderCache
Change-Id: Ia936ff8bb639a197b0b3a8e07c97a66edd57dd10
This reused the gender state of a user on a page. This is helpful for
special pages which shows the group name, because the each group name
used gender, which result in often use.
Change-Id: I8e816f54aaa100c3333e84e19299fd194323341d
This allows HTCP purges for different domains to be sent to different
multicast groups, based on regexes. Mark requested this so we could
separate the multicast groups for upload caches and text caches.
This code is UNTESTED, I'm mostly submitting this as a proof of concept
and to invite review by other core devs (specifically Tim).
Change-Id: Ie333a04131d6ca8394884ed1054f2baff55ab2d1
This has to be done in 3 places, because there are 3 public entry
points.
I originally submitted this to fight duplicates I thought I was seeing
in production, but it turns out I'm blind and the URLs weren't
duplicates after all. Nevertheless, preventing duplicate purges in the
SquidUpdate class is a good idea.
Change-Id: Idc21dd7d0b3b79572853b787fac746454d9178ea
* Tweaked filecache fallback in fileCachedPage() to try the raw title param. If the DB is down, we can get most views of titles with colons in them to work this way. Previously, it could fail on an interwiki lookup.
* Use default .cache extension for ResourceFileCache. No need for js/css extension, and makes extension sanity check in prune script simpler.
* Removed redundant setting of mExt in ObjectFileCache
* Added FileCacheBase::*MissesRecent() functions for counting cache misses from different visitors.
* Made ObjectFileCache more generic.
* Cleaned up FileCacheBase::checkCacheDirs().
* Added FileCacheBase::typeSubdirectory() function and overwrote in HTMLFileCache. Fixes r98405 invalidating all existing cache due to directory change.
* Simplified FileCacheBase::checkCacheDirs() a bit
ResourceLoader:
* Use ResourceFileCache to handle load() requests, if $wgUseFileCache. Only caches requests for default language and skins. Single modules requests are always cached, whereas others require a certain threshold of traffic.
* Added ResourceFileCache class (functionality was initially to be in ObjectFileCache).
* Rewrote class and split into three classes: a base class, and html cache and a more generic cache to be used later.
* The new classes now use RequestContext.
* Renamed fetchPageText() -> fetchText().
* Split out new saveText() function from saveToFileCache().
* Various other cleanups and fixes.
Also fixed backwards setting of $wgDisableCounters in rebuildFileCache.php.
This may be related to things failing on load from External Storage; however we have not yet been able to verify this.
Tweaks MessageCache::loadFromDB() and MessageCache::getMsgFromNamespace() to avoid storing empty cache entries when loading text fails.
When building initial cache if we get a failure we'll log and store a '!TOO BIG' message which requests on-demand load later.
If an on-demand load failures, we'll log and return the false through but won't update the cache with the bad value.
To enable the logging in production, set up a $wgDebugLogFiles entry for 'MessageCache'.
Note that MessageCache::loadFromDB() bypasses Revision's text entry memcaching and may cause a lot of ES fetches at once.
However any ES failures *should* already be logged in the 'ExternalStoreDB' log file.
$wgDeferredUpdateList[] = new HTMLCacheUpdate( $title, "sometable" );
I kept b/c with $wgDeferredUpdateList for now, but seeing as only 3 exts in svn use it (FileSearch, FlaggedRevs and WikiScripts), I'd like to deprecate it pretty soon :)
$wgContLang in CategoryPage was probably removed with the better directionality improvements.
$wgServer, $wgRequest in getFullUrl() unused since r94375
Other variables added in r95396 iwtransclusion merge.
Breaks unit tests as below, not going to be able to fix them before I disappear for the evening, so might aswell leave trunk clean
ArticleTablesTest testbug14404
Error:
ArticleTablesTest::testbug14404
Undefined offset: 0
/home/ci/cruisecontrol-bin-2.8.3/projects/mw/source/tests/phpunit/includes/ArticleTablesTest.php:31
/home/ci/cruisecontrol-bin-2.8.3/projects/mw/source/tests/phpunit/MediaWikiTestCase.php:60
/home/ci/cruisecontrol-bin-2.8.3/projects/mw/source/tests/phpunit/MediaWikiPHPUnitCommand.php:20
/home/ci/cruisecontrol-bin-2.8.3/projects/mw/source/tests/phpunit/phpunit.php:60
ParserTests testParserTest #552 - testParserTest with data set #551
Failure:
ParserTests::testParserTest with data set #551 ('RAW magic word', '{{RAW:QUERTY}}', '<p><a href="/index.php?title=Template:QUERTY&action=edit&redlink=1" class="new" title="Template:QUERTY (page does not exist)">Template:QUERTY</a>
</p>', '', '')
RAW magic word
Failed asserting that two strings are equal.
--- Expected
+++ Actual
@@ @@
-<p><a href="/index.php?title=Template:QUERTY&action=edit&redlink=1" class="new" title="Template:QUERTY (page does not exist)">Template:QUERTY</a>
+<p><a href="/index.php?title=Template:RAW:QUERTY&action=edit&redlink=1" class="new" title="Template:RAW:QUERTY (page does not exist)">Template:RAW:QUERTY</a>
</p>
/home/ci/cruisecontrol-bin-2.8.3/projects/mw/source/tests/phpunit/includes/parser/NewParserTest.php:545
/home/ci/cruisecontrol-bin-2.8.3/projects/mw/source/tests/phpunit/MediaWikiTestCase.php:60
/home/ci/cruisecontrol-bin-2.8.3/projects/mw/source/tests/phpunit/MediaWikiPHPUnitCommand.php:20
/home/ci/cruisecontrol-bin-2.8.3/projects/mw/source/tests/phpunit/phpunit.php:60
** Tweaked addAutopromoteOnceGroups() calls for performance
** Some doc tweaks and fixes
* Added NS_SPECIAL short-circuit on $wgTitle for file cache and removed unnecessary "is null" check
* Pushed "$action != 'raw'" check into HTMLFileCache
* Removed useless return value from performRequest()
* Added type hint to performAction()
