Merge lp:~abychko/percona-server/bug1050654-5.6 into lp:percona-server/5.6

[Alexey Bychko]

bug 1050654, fix for 5.6
[+] additional merge from bug 1165098, to get build files with correct bzr ids. last time some part of merge from 5.5 was lost while fighting additional and unexpected file-ids issue
[+] added Fedora linker script to the build to split symbols by versions
[+] added 18_1 symbols for 5.6
[+] added symbols for compatibility
php build and php bundled tests are ok
#####################################
this commit is port from Fedora with small changes for Percona Server v5.6

[Alexey Kopytov (akopytov)]

Alexey,

I don't see changes to mysqladmin.cc we discussed last week. I.e. it still uses the undocumented make_() API calls, though with your patch they are not even exported?

review: Needs Information

[Alexey Bychko (abychko)]

Alexey, changes are in build/rpm/mysql-dubious-exports.patch, similar to 5.5.
may be better to finish that patch for 5.5 and 5.6?

[Alexey Bychko (abychko)]

LP diff can't be generated, for some unknown reason. seems branch should be reviewed by looking at the commit directly

message from LP:

Launchpad encountered an internal error during the following operation:
generating the diff for a merge proposal. It was logged with id
OOPS-d50983ee4df7d5cda50c9990c5425537. Sorry for the inconvenience.

[Alexey Kopytov (akopytov)]

I did review the revision in a local clone of the lp:~abychko/percona-server/bug1050654-5.6 branch:

$ bzr pull --overwrite
Using saved parent location: bzr+ssh://bazaar.launchpad.net/~abychko/percona-server/bug1050654-5.6/
No revisions or tags to pull.
$ bzr diff -c-1 | grep sha1
$

[Alexey Kopytov (akopytov)]

But, but, but... I still don't see the discussed changes in the diff? I also doublechecked with a local clone.

review: Needs Fixing

[Alexey Bychko (abychko)]

http://bazaar.launchpad.net/~abychko/percona-server/bug1050654-5.6/revision/337/build/rpm/mysql-dubious-exports.patch line #20

[Alexey Kopytov (akopytov)]

Indeed, sorry. This was partially LP's fault with showing a wrong diff, and my fault with checking the code rather than the patch.

review: Approve

=== added file 'Percona-Server/libmysql/libmysql.map'
--- Percona-Server/libmysql/libmysql.map	1970-01-01 00:00:00 +0000
+++ Percona-Server/libmysql/libmysql.map	2013-04-15 15:09:43 +0000
@@ -0,0 +1,155 @@
+# symbols exported from mysql 5.1
+libmysqlclient_16 {
+  global:
+	_fini;
+	_init;
+	my_init;
+	myodbc_remove_escape;
+	mysql_affected_rows;
+	mysql_autocommit;
+	mysql_change_user;
+	mysql_character_set_name;
+	mysql_close;
+	mysql_commit;
+	mysql_data_seek;
+	mysql_debug;
+	mysql_dump_debug_info;
+	mysql_embedded;
+	mysql_eof;
+	mysql_errno;
+	mysql_error;
+	mysql_escape_string;
+	mysql_fetch_field;
+	mysql_fetch_field_direct;
+	mysql_fetch_fields;
+	mysql_fetch_lengths;
+	mysql_fetch_row;
+	mysql_field_count;
+	mysql_field_seek;
+	mysql_field_tell;
+	mysql_free_result;
+	mysql_get_character_set_info;
+	mysql_get_client_info;
+	mysql_get_client_version;
+	mysql_get_host_info;
+	mysql_get_parameters;
+	mysql_get_proto_info;
+	mysql_get_server_info;
+	mysql_get_server_version;
+	mysql_get_ssl_cipher;
+	mysql_hex_string;
+	mysql_info;
+	mysql_init;
+	mysql_insert_id;
+	mysql_kill;
+	mysql_list_dbs;
+	mysql_list_fields;
+	mysql_list_processes;
+	mysql_list_tables;
+	mysql_more_results;
+	mysql_next_result;
+	mysql_num_fields;
+	mysql_num_rows;
+	mysql_options;
+	mysql_ping;
+	mysql_query;
+	mysql_read_query_result;
+	mysql_real_connect;
+	mysql_real_escape_string;
+	mysql_real_query;
+	mysql_refresh;
+	mysql_rollback;
+	mysql_row_seek;
+	mysql_row_tell;
+	mysql_select_db;
+	mysql_send_query;
+	mysql_server_end;
+	mysql_server_init;
+	mysql_set_character_set;
+	mysql_set_local_infile_default;
+	mysql_set_local_infile_handler;
+	mysql_set_server_option;
+	mysql_shutdown;
+	mysql_sqlstate;
+	mysql_ssl_set;
+	mysql_stat;
+	mysql_stmt_affected_rows;
+	mysql_stmt_attr_get;
+	mysql_stmt_attr_set;
+	mysql_stmt_bind_param;
+	mysql_stmt_bind_result;
+	mysql_stmt_close;
+	mysql_stmt_data_seek;
+	mysql_stmt_errno;
+	mysql_stmt_error;
+	mysql_stmt_execute;
+	mysql_stmt_fetch;
+	mysql_stmt_fetch_column;
+	mysql_stmt_field_count;
+	mysql_stmt_free_result;
+	mysql_stmt_init;
+	mysql_stmt_insert_id;
+	mysql_stmt_num_rows;
+	mysql_stmt_param_count;
+	mysql_stmt_param_metadata;
+	mysql_stmt_prepare;
+	mysql_stmt_reset;
+	mysql_stmt_result_metadata;
+	mysql_stmt_row_seek;
+	mysql_stmt_row_tell;
+	mysql_stmt_send_long_data;
+	mysql_stmt_sqlstate;
+	mysql_stmt_store_result;
+	mysql_store_result;
+	mysql_thread_end;
+	mysql_thread_id;
+	mysql_thread_init;
+	mysql_thread_safe;
+	mysql_use_result;
+	mysql_warning_count;
+# These are documented in Paul DuBois' MySQL book, so we treat them as part
+# of the de-facto API.
+	free_defaults;
+	handle_options;
+	load_defaults;
+	my_print_help;
+# This isn't really documented anywhere, but it seems to be part of the
+# de-facto API as well.  We're not going to export the deprecated version
+# make_scrambled_password, however.
+	my_make_scrambled_password;
+# This really shouldn't be exported, but some applications use it as a
+# workaround for inadequate threading support; see bug #846602
+	THR_KEY_mysys;
+  local:
+	*;
+};
+# symbols added in mysql 5.5
+libmysqlclient_18 {
+  global:
+	mysql_client_find_plugin;
+	mysql_client_register_plugin;
+	mysql_load_plugin;
+	mysql_load_plugin_v;
+	mysql_plugin_options;
+	mysql_stmt_next_result;
+#
+# Ideally the following symbols wouldn't be exported, but various applications
+# require them.  We limit the namespace damage by prefixing mysql_
+# (see mysql-dubious-exports.patch), which means the symbols are not present
+# in libmysqlclient_16.
+#
+# mysql-connector-odbc requires these
+	mysql_default_charset_info;
+	mysql_get_charset;
+	mysql_get_charset_by_csname;
+	mysql_net_realloc;
+# PHP's mysqli.so requires this (via the ER() macro)
+	mysql_client_errors;
+# mydumper requires this (added by remi)
+	my_net_read;
+};
+# symbols added in mysql 5.6
+libmysqlclient_18_1 {
+  global:
+	get_tty_password;
+};