On 12/16/2016 09:31 PM, Sergei Golubchik wrote:

Hi, Alexander!

Looks ok. One question below

On Dec 15, Alexander Barkov wrote:

...

commit 72817bb386320080de87f9cfe899e6f74341d251 Author: Alexander Barkov Date: Thu Dec 15 19:40:09 2016 +0400

MDEV-10524 Assertion `arg1_int >= 0' failed in Item_func_additive_op::result_precision()

This change is a backport from 10.0 to 5.5 for: 1. The full patch for: MDEV-4841 Wrong character set of ADDTIME() and DATE_ADD() 9adb6e991ec87b65d04929f115d9d0c899e4ab19

2. A small fragment of: MDEV-5298 Illegal mix of collations on timestamp 03f6778d61a74bdd7d09103a16473a2a5624cf66 which overrides Item_temporal_hybrid_func::cmp_type(), and adds a new line into cache_temporal_4265.result.

diff --git a/sql/item_func.h b/sql/item_func.h index 667be3c..cfa97ea 100644 --- a/sql/item_func.h +++ b/sql/item_func.h @@ -46,6 +47,8 @@ class Item_func :public Item_result_field void count_datetime_length(Item **item, uint nitems); bool count_string_result_length(enum_field_types field_type, Item **item, uint nitems); + /* maybe_null can't be changed by parameters or used table state */ + bool persistent_maybe_null;

What's this? it didn't come from 10.0, as far as I can see. Nor is it used anywhere in the patch.

Oops. It's not needed. Thanks for noticing this. Pushed. Thanks for review!

...

public: uint arg_count; table_map used_tables_cache, not_null_tables_cache;

Regards, Sergei Chief Architect MariaDB and security@mariadb.org