Hi Sergei, On Mon, Mar 24, 2014 at 12:25:20PM +0100, Sergei Golubchik wrote:
Hi, Sergey!
On Mar 24, Sergey Vojtovich wrote:
This is second list of questions I came up while merging test cases: - --ignore-builtin-innodb is deprecated in 5.6, do we want to deprecate it too?
No, of course not. It was introduced in 5.1 so that one could load the engine from storage/innodb_plugin instead of the compiled-in engine from storage/innobase. The tree had two versions of innodb.
Now MySQL has only one version of InnoDB, this switch can only be used to load XtraDB instead.
But MariaDB is shipped with both XtraDB and InnoDB, and we want our users to have a choice, they need --ignore-builtin-innodb. I thought they implemented a generic way to disable builtin plugins as a replacement, but apparently they didn't. Ok.
- LOW_PRIORITY WRITE is deprecate in 5.6, do we want to deprecate it too?
Yes, I think so. LOCK TABLE ... LOW_PRIORITY WRITE is exactly identical to LOCK TABLE ... WRITE, "LOW_PRIORITY" is ignored basically since MDLs were introduced.
Will create a task.
The following items seem to be not merged to MariaDB. Do we want to merge them? WL#5331 Support Unicode for Windows command line client
MDEV-4118
Bug#12688860 : SECURITY RECOMMENDATION: PASSWORDS ON CLI
I don't know. Looks annoying to me. And it doesn't help anyway - remember how many bug reports we've got like "MYSQLD DOESN'T START, IT ONLY PRINTS `Fatal error: Please read Security section of the manual to find out how to run mysqld as root!' WHAT SHOULD I DO?"
Frankly speaking I'm uncertain about this fix too. Let's leave it unless anybody objects.
WL#5855 (EXPLAIN FORMAT=JSON) partition test
what do you mean?
It was copied from partition.test. A verbatim worklog name seems to be: WL#5855: Print EXPLAIN in JSON format Which we're missing anyway.
WL#5980 - Allow remote tablespaces (need to check)
it's a support for DATA DIRECTORY in InnoDB please create a jira issue for 10.1.
Ok.
WL5602 Providing password hashing with appropriate strength
No, that's badly implemented support for SHA256 password hashing. We might want to do it - properly - if there will be a convincing reason to.
Ok.
Bug#12316645: Wrong cost calculation with optimizer_join_cache_level settings
It comes with test cases. Do they pass? Fail? I suppose you can create a Jira issue for that, like you did for other bugs.
I think I found it while analyzing diff like this: --- mysql-test/r/count_distinct.result 2010-04-13 09:38:28 +0000 +++ mysql-test/r/count_distinct.result 2011-06-08 13:24:56 +0000 @@ -49,7 +49,7 @@ isbn city @bar:=t1.libname a 000 New York New York Public Libra 2 SELECT @bar; @bar -Berkeley Public2 +NYC Lib select t2.isbn,city,concat(@bar:=t1.libname),count(distinct t1.libname) as a from t3 left join t1 on t3.libname=t1.libname left join t2 on t3.isbn=t2.isbn group by city having count(distinct @@ -59,7 +59,7 @@ isbn city concat(@bar:=t1.libname) a 000 New York New York Public Libra 2 SELECT @bar; @bar -Berkeley Public2 +NYC Lib drop table t1, t2, t3; create table t1 (f1 int); insert into t1 values (1); optimizer_join_cache_level was removed in 5.6.3 in favour of optimizer_switch. I'm not sure if there is corresponding 10.0 option affected.
WL#5602 - SHA256 password authentication
See above.
Bugs#12635232: VALGRIND WARNINGS: IS_IPV6, IS_IPV4, INET6_ATON, INET6_NTOA + MULTIBYTE CHARSET Bug#14040277 UNINITIALIZED VALUE REFERENCED IN STR_TO_IPV6 And a lot of other IPV6 related stuff.
Please create a jira task for 10.1 (adding ipv6/ipv4 functions like in 5.6). Unless we have it already, there are few ipv6-related tasks.
There are: MDEV-274 - The data type for IPv6/IPv4 addresses in MariaDB MDEV-4051 - INET6_ATON() and INET6_NTOA() But there is nothing like IS_IPV6 etc. I'll create a task. Regards, Sergey