From: kevin.lewis
Date: May 9 2012 11:22pm
Subject: bzr push into mysql-trunk branch (kevin.lewis:3808 to 3809)
List-Archive: http://lists.mysql.com/commits/144180
Message-Id: <20120509232238.5445C217263A@dhcp-adc-twvpn-2-vpnpool-10-154-54-88.vpn.oracle.com>
MIME-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Content-Transfer-Encoding: 7bit

 3809 kevin.lewis@stripped	2012-05-09
      5980 patch - Fixed some comments to address the previous TODO comments.
      Another fix for keeping the crashing test from running on non-debug platforms.

    modified:
      mysql-test/suite/innodb/t/innodb-multiple-tablespaces.test
      storage/innobase/fil/fil0fil.cc
 3808 kevin.lewis@stripped	2012-05-09
      Prevent crashing test from running where it should not.

    modified:
      mysql-test/suite/innodb/t/innodb-multiple-tablespaces.test
=== modified file 'mysql-test/suite/innodb/t/innodb-multiple-tablespaces.test'
--- a/mysql-test/suite/innodb/t/innodb-multiple-tablespaces.test	revid:kevin.lewis@stripped
+++ b/mysql-test/suite/innodb/t/innodb-multiple-tablespaces.test	revid:kevin.lewis@stripped
@@ -4,6 +4,9 @@
 
 --source include/have_innodb.inc
 
+# Must have debug code to use SET SESSION debug
+--source include/have_debug.inc
+
 # Valgrind would complain about memory leaks when we crash on purpose.
 --source include/not_valgrind.inc
 

=== modified file 'storage/innobase/fil/fil0fil.cc'
--- a/storage/innobase/fil/fil0fil.cc	revid:kevin.lewis@stripped
+++ b/storage/innobase/fil/fil0fil.cc	revid:kevin.lewis@stripped
@@ -1875,7 +1875,6 @@ fil_write_flushed_lsn_to_data_files(
 		and the UNDO log tablespaces (if separate) are always open. */
 
 		if (space->purpose == FIL_TABLESPACE) {
-//		    && !fil_is_user_tablespace_id(space->id)) {   // KLTEST
 
 			ulint	sum_of_sizes = 0;
 
@@ -2151,6 +2150,12 @@ created does not exist, then we create t
 
 Note that ibbackup --apply-log sets fil_path_to_mysql_datadir to point to the
 datadir that we should use in replaying the file operations.
+
+InnoDB recovery does not replay these fully since it always sets the space id
+to zero. But ibbackup does replay them.  TODO: If remote tablespaces are used,
+ibbackup will only create tables in the default directory since MLOG_FILE_CREATE
+and MLOG_FILE_CREATE2 only know the tablename, not the path.
+
 @return end of log record, or NULL if the record was not completely
 contained between ptr and end_ptr */
 UNIV_INTERN
@@ -2268,8 +2273,7 @@ fil_op_log_parse_or_replay(
 			if (fil_get_space_id_for_table(new_name)
 			    == ULINT_UNDEFINED) {
 				/* We do not care about the old name, that
-				is why we pass NULL as the first argument.
-				TODO: If file is remote, we do not know the path */
+				is why we pass NULL as the first argument. */
 				if (!fil_rename_tablespace(NULL, space_id,
 							   new_name, NULL)) {
 					ut_error;
@@ -3015,9 +3019,6 @@ directory of a running mysqld program. W
 path '.'. Tables created with CREATE TEMPORARY TABLE we place in the temp
 dir of the mysqld server.
 
-TODO: If this is being called during recovery replaying an MLOG_FILE_CREATE
-or MLOG_FILE_CREATE2 record and the DATA DICTIONARY path was used, then we
-do not have a tablename that looks like database/tablename.
 @return	DB_SUCCESS or error code */
 UNIV_INTERN
 dberr_t

No bundle (reason: useless for push emails).