Further categorylinks schema changes

Per review by Tim, I made two changes:

1) Fix cl_sortkey to be varbinary(255).

2) Expand cl_collation to varbinary(32), and change $wgCollationVersion
to $wgCategoryCollation, to account for the variety of collations we
might have.  tinyint is too small.  I could have gone with int, but
that's annoyingly inscrutable in practice, as we all know from namespace
fields.

To make the upgrade easier for non-trunk users, I updated the old patch
file to incorporate the new changes, using the updatelog table so that
people upgrading from 1.16 won't have to do two alters on categorylinks.
I didn't test the upgrade-from-1.16 code path yet, so if anyone tests
that and it seems not to break, commenting to that effect would be
appreciated.

Also removed wfDeprecated() from archive().  Do *not* add this to
functions that are still actively used in core.  If you think this
function is so terrible that it really mustn't be used, remove callers
yourself, don't pester every single developer with messages in the hope
that someone else will do it for you.

diff --git a/includes/DefaultSettings.php b/includes/DefaultSettings.php
index 333b3a5..aa38e90 100644 (file)
--- a/includes/DefaultSettings.php
+++ b/includes/DefaultSettings.php
@@ -4449,10 +4449,13 @@ $wgCategoryPagingLimit = 200;
 /**
  * A version indicator for collations that will be stored in cl_collation for
  * all new rows.  Used when the collation algorithm changes: a script checks
- * for all rows where cl_collation != $wgCollationVersion and regenerates
+ * for all rows where cl_collation != $wgCategoryCollation and regenerates
  * cl_sortkey based on the page name and cl_sortkey_prefix.
+ *
+ * Currently only supports 'uppercase', which just uppercases the string.  This
+ * is a dummy collation, to be replaced later by real ones.
  */
-$wgCollationVersion = 1;
+$wgCategoryCollation = 'uppercase';
 
 /** @} */ # End categories }
 

diff --git a/includes/LinksUpdate.php b/includes/LinksUpdate.php
index 7092d3d..4892ff0 100644 (file)
--- a/includes/LinksUpdate.php
+++ b/includes/LinksUpdate.php
@@ -426,7 +426,7 @@ class LinksUpdate {
         * @private
         */
        function getCategoryInsertions( $existing = array() ) {
-               global $wgContLang, $wgCollationVersion;
+               global $wgContLang, $wgCategoryCollation;
                $diffs = array_diff_assoc( $this->mCategories, $existing );
                $arr = array();
                foreach ( $diffs as $name => $sortkey ) {
@@ -465,7 +465,7 @@ class LinksUpdate {
                                'cl_sortkey' => $sortkey,
                                'cl_timestamp' => $this->mDb->timestamp(),
                                'cl_sortkey_prefix' => $prefix,
-                               'cl_collation' => $wgCollationVersion,
+                               'cl_collation' => $wgCategoryCollation,
                                'cl_type' => $type,
                        );
                }

diff --git a/includes/installer/MysqlUpdater.php b/includes/installer/MysqlUpdater.php
index 7d76860..ecd1242 100644 (file)
--- a/includes/installer/MysqlUpdater.php
+++ b/includes/installer/MysqlUpdater.php
@@ -165,6 +165,7 @@ class MysqlUpdater extends DatabaseUpdater {
                        array( 'drop_index_if_exists', 'iwlinks', 'iwl_prefix',  'patch-kill-iwl_prefix.sql' ),
                        array( 'drop_index_if_exists', 'iwlinks', 'iwl_prefix_from_title', 'patch-kill-iwl_pft.sql' ),
                        array( 'addField', 'categorylinks', 'cl_collation', 'patch-categorylinks-better-collation.sql' ),
+                       array( 'do_cl_fields_update' ),
                        array( 'do_collation_update' ),
                );
        }

diff --git a/maintenance/archives/patch-categorylinks-better-collation.sql b/maintenance/archives/patch-categorylinks-better-collation.sql
index 844ad9c..7c711a3 100644 (file)
--- a/maintenance/archives/patch-categorylinks-better-collation.sql
+++ b/maintenance/archives/patch-categorylinks-better-collation.sql
@@ -1,11 +1,15 @@
 --
 -- patch-categorylinks-better-collation.sql
 --
--- Bugs 164, 1211, 23682.
+-- Bugs 164, 1211, 23682.  This is the second version of this patch; the
+-- changes are also incorporated into patch-categorylinks-better-collation2.sql,
+-- for the benefit of trunk users who applied the original.
 ALTER TABLE /*$wgDBprefix*/categorylinks
+       CHANGE COLUMN cl_sortkey cl_sortkey varbinary(255) NOT NULL default '',
        ADD COLUMN cl_sortkey_prefix varchar(255) binary NOT NULL default '',
-       ADD COLUMN cl_collation tinyint NOT NULL default 0,
+       ADD COLUMN cl_collation varbinary(32) NOT NULL default '',
        ADD COLUMN cl_type ENUM('page', 'subcat', 'file') NOT NULL default 'page',
        ADD INDEX (cl_collation),
        DROP INDEX cl_sortkey,
        ADD INDEX cl_sortkey (cl_to, cl_type, cl_sortkey, cl_from);
+INSERT IGNORE INTO /*$wgDBprefix*/updatelog (ul_key) VALUES ('cl_fields_update');

diff --git a/maintenance/archives/patch-categorylinks-better-collation2.sql b/maintenance/archives/patch-categorylinks-better-collation2.sql
new file mode 100644 (file)
index 0000000..7515c56
--- /dev/null
+++ b/maintenance/archives/patch-categorylinks-better-collation2.sql
@@ -0,0 +1,12 @@
+--
+-- patch-categorylinks-better-collation2.sql
+--
+-- Bugs 164, 1211, 23682.  This patch exists for trunk users who already
+-- applied the first patch in its original version.  The first patch was
+-- updated to incorporate the changes as well, so as not to do two alters on a
+-- large table unnecessarily for people upgrading from 1.16, so this will be
+-- skipped if unneeded.
+ALTER TABLE /*$wgDBprefix*/categorylinks
+       CHANGE COLUMN cl_sortkey cl_sortkey varbinary(255) NOT NULL default '',
+       CHANGE COLUMN cl_collation cl_collation varbinary(32) NOT NULL default '';
+INSERT IGNORE INTO /*$wgDBprefix*/updatelog (ul_key) VALUES ('cl_fields_update');

diff --git a/maintenance/tables.sql b/maintenance/tables.sql
index 6f303d3..dcb776d 100644 (file)
--- a/maintenance/tables.sql
+++ b/maintenance/tables.sql
@@ -493,12 +493,7 @@ CREATE TABLE /*_*/categorylinks (
   -- A binary string obtained by applying a sortkey generation algorithm
   -- (Language::convertToSortkey()) to page_title, or cl_sortkey_prefix . "\0"
   -- . page_title if cl_sortkey_prefix is nonempty.
-  --
-  -- Truncate so that the cl_sortkey key fits in 1000 bytes (MyISAM 5 with
-  -- server_character_set=utf8).  FIXME: this truncation probably makes no
-  -- sense anymore; we should be using varbinary for this, utf8 will break
-  -- everything.
-  cl_sortkey varchar(70) binary NOT NULL default '',
+  cl_sortkey varbinary(255) NOT NULL default '',
 
   -- A prefix for the raw sortkey manually specified by the user, either via
   -- [[Category:Foo|prefix]] or {{defaultsort:prefix}}.  If nonempty, it's
@@ -511,12 +506,12 @@ CREATE TABLE /*_*/categorylinks (
   -- sorting method by approximate addition time.
   cl_timestamp timestamp NOT NULL,
 
-  -- Stores $wgCollationVersion at the time cl_sortkey was generated.  This can
-  -- be used to install new collation versions, tracking which rows are not yet
-  -- updated.  0 means no collation, this is a legacy row that needs to be
+  -- Stores $wgCategoryCollation at the time cl_sortkey was generated.  This
+  -- can be used to install new collation versions, tracking which rows are not
+  -- yet updated.  '' means no collation, this is a legacy row that needs to be
   -- updated by updateCollation.php.  In the future, it might be possible to
   -- specify different collations per category.
-  cl_collation tinyint NOT NULL default 0,
+  cl_collation varbinary(32) NOT NULL default '',
 
   -- Stores whether cl_from is a category, file, or other page, so we can
   -- paginate the three categories separately.  This never has to be updated

diff --git a/maintenance/updateCollation.php b/maintenance/updateCollation.php
index 52b31be..7b91660 100644 (file)
--- a/maintenance/updateCollation.php
+++ b/maintenance/updateCollation.php
@@ -15,10 +15,10 @@ class UpdateCollation extends Maintenance {
        public function __construct() {
                parent::__construct();
 
-               global $wgCollationVersion;
+               global $wgCategoryCollation;
                $this->mDescription = <<<TEXT
 This script will find all rows in the categorylinks table whose collation is
-out-of-date (cl_collation != $wgCollationVersion) and repopulate cl_sortkey
+out-of-date (cl_collation != '$wgCategoryCollation') and repopulate cl_sortkey
 using the page title and cl_sortkey_prefix.  If everything's collation is
 up-to-date, it will do nothing.
 TEXT;
@@ -27,13 +27,13 @@ TEXT;
        }
        
        public function execute() {
-               global $wgCollationVersion, $wgContLang;
+               global $wgCategoryCollation, $wgContLang;
 
                $dbw = wfGetDB( DB_MASTER );
                $count = $dbw->selectField(
                        'categorylinks',
                        'COUNT(*)',
-                       'cl_collation != ' . $dbw->addQuotes( $wgCollationVersion ),
+                       'cl_collation != ' . $dbw->addQuotes( $wgCategoryCollation ),
                        __METHOD__
                );
 
@@ -51,7 +51,7 @@ TEXT;
                                        'cl_sortkey', 'page_namespace', 'page_title'
                                ),
                                array(
-                                       'cl_collation != ' . $dbw->addQuotes( $wgCollationVersion ),
+                                       'cl_collation != ' . $dbw->addQuotes( $wgCategoryCollation ),
                                        'cl_from = page_id'
                                ),
                                __METHOD__,
@@ -89,7 +89,7 @@ TEXT;
                                                'cl_sortkey' => $wgContLang->convertToSortkey(
                                                        $title->getCategorySortkey( $prefix ) ),
                                                'cl_sortkey_prefix' => $prefix,
-                                               'cl_collation' => $wgCollationVersion,
+                                               'cl_collation' => $wgCategoryCollation,
                                                'cl_type' => $type,
                                                'cl_timestamp = cl_timestamp',
                                        ),

diff --git a/maintenance/updaters.inc b/maintenance/updaters.inc
index 856c3e7..ab99040 100644 (file)
--- a/maintenance/updaters.inc
+++ b/maintenance/updaters.inc
@@ -103,7 +103,6 @@ function do_all_updates( $shared = false, $purge = true ) {
 }
 
 function archive( $name ) {
-       wfDeprecated( __FUNCTION__ );
        return DatabaseBase::patchPath( $name );
 }
 
@@ -833,12 +832,23 @@ function do_populate_rev_len() {
        $task->execute();
 }
 
+function do_cl_fields_update() {
+       if ( update_row_exists( 'cl_fields_update' ) ) {
+               wfOut( "...categorylinks up-to-date.\n" );
+               return;
+       }
+       wfOut( 'Updating categorylinks (again)...' );
+       global $wgDatabase;
+       $wgDatabase->sourceFile( archive( 'patch-categorylinks-better-collation2.sql' ) );
+       wfOut( "done.\n" );
+}
+
 function do_collation_update() {
-       global $wgDatabase, $wgCollationVersion;
+       global $wgDatabase, $wgCategoryCollation;
        if ( $wgDatabase->selectField(
                'categorylinks',
                'COUNT(*)',
-               'cl_collation != ' . $wgDatabase->addQuotes( $wgCollationVersion ),
+               'cl_collation != ' . $wgDatabase->addQuotes( $wgCategoryCollation ),
                __FUNCTION__
        ) == 0 ) {
                wfOut( "...collations up-to-date.\n" );