Hi Sergei,
Please review a patch that merges a new read-only variables
--explicit-defaults-for-timestamp from MySQL.
Note, the original MySQL patch updates many tests,
to make the tests independent from the compiled-in default
--explicit-defaults-for-timestamp value. I guess Oracle is planning to
change the default value to ON eventually.
I merged these changes for tests as well.
As a result, all tests pass in both "mtr" and
"mtr --mysqld=--explicit-defaults-for-timestamp",
which I think is good.
In addition to the original MySQL changes in mtr, I also added
two tests that describe specific behavior:
sys_vars.explicit_defaults_for_timestamp_off
sys_vars.explicit_defaults_for_timestamp_on
Note, the patch assumes that the server and the slave have the same
--explicit-defaults-for-timeatamp. Monty suggested to make this variable
dynamic and replication-safe. We added a separate task for this:
MDEV-8455 Make --explicit-defaults-for-timestamp dynamic
I plan to do it in the next 10.1 sprint round.
Thanks.
