* Fix up messages
* For new FileRepo::append(), use flags to determine whether to delete or not
* Add more error checking for appending
* Fix a couple of places in Revision.php and LogPage.php where DB errors were produced when comment was null
* Remove bogus checking for !$comment, etc on the DONE phase of chunked uploading
* Don't pretend to return a value when raising an exception
* Add more tests for chunked uploads
* Verify that Status::getErrorsArray() (at least where it is used in ApiUpload::execute()) returns an array that we can pass to dieUsageMessage()
* Ensure that checkWarnings(), etc work only on the complete file
Updated all nextSequenceValue calls with new sequence names.
OverlordQ stated he'll handle changes to Postgres scripts.
Need someone to change DB2 scripts.
* Refactored to remove massive duplication
* Removed page parameter and associated contextPage member variable, doesn't seem to do anything.
* Put ID lists into a single ids parameter and member variable instead of having one for each type.
* Fixed inappropriate call of Title::newFromUrl(), always wrong
* Don't pretend to use the return value from functions that don't return anything, this reduces readability.
* Use the table names for deleteKey/typeName values, they look more like English
* Use protected not private
* Remove requirement for log type to be specified in the target
* Use POST for RevisionDelete entry forms, to avoid URL size limits and issues with non-PATH_INFO URLs
* Don't purge all pages that use the given file
* LocalFile::purgeCache() already calls purgeHistory,() no need to do it again. But do call purgeDescription().
* Removed token from unhide=1 links, unnecessary
* Tokens are necessary on file streaming links, added them
* Fixed private data leakage due to incorrect use of LocalRepo::newFromArchiveName(). Non-existent placeholder file was returned which meant that $oimage->userCan(File::DELETED_FILE) was always true. Pass the archive name to tryShowFile() instead of the storage key.
* Using ls_field='oi_timestamp' is not correct, oi_timestamp refers to the timestamp of the revision in question, whereas the key that is stored is the timestamp of the previous revision (i.e. the timestamp in oi_archive_name). oi_archive_name would be more correct, although only half the field is used.
Elsewhere:
* Added missing message filehist-missing
* Fixed double asterisk in Status::getWikiText()
* Fixed escaping of the target parameter to Special:RevisionDelete from ImagePage
* Deleted FileStore.php. Deprecated in filerepo refactor except for get()/export() but somehow resurrected by RevisionDelete. Hopefully this will be the end of it. New interfaces will be added for wfStreamFile() in a later commit.
* Added convenience function File::getStorageKey(), factored out of Special:Undelete
* Added convenience function Revision::newFromArchiveRow(), factored out of Special:Undelete and Special:RevisionDelete
* Fixed notice in Special:Upload, uninitialised $pageText
FIXME: current revision can be suppressed on undeletion causing an unauthenticated unsuppress. Comments indicate this is a known issue. I fixed the parser cache pollution in this case but not the rest.
* Fix broken recursion guard in LoadBalancer::reportConnectionError(), which was causing getConnection() to return false on the second and subsequent errors, instead of throwing an exception. Revert incorrect fix r41229/r41230.
* Boolean parameters are widely accepted to reduce readability. Replaced the new boolean parameters with class constant parameters instead.
* Re-added Revision::revText(), for backwards compatibility
* The getUser()/getUserText() changes near line 1223 of SpecialUndelete.php were incorrect, $file is an ArchivedFile not a Revision, and doesn't have any $isPublic parameters.
* On read, spend less time checking on dead slaves
* Add randomInsert() to ES. This does the cluster picking for us
* Make revision text use randomInsert(). On write, fails-over to other clusters as needed instead of throwing db errors
* Don't pass $data or $flags by reference. They're never accessed again, so it makes no sense.
* Document in hooks.txt.
The interface for this hook seems very messy. It seems like it would be much cleaner to expose $data as $url (null if stored locally); and $flags as an array, or something like that. If the text is gzipped, the text can be extracted from the Revision and re-gzipped if you really want to for some strange reason.
