Sergei Golubchik

Yes, of course. Please, do push and request a review. Also, if you think some of my suggestions should not be implemented (for whatever reason), please say so in a...

I'm sorry, I didn't understand. Are you requesting a review or do you still want to change something before you want me to take a look? As for moving path...

A question (please, don't change anything yet, it might make the review obsolete before I even finish): why did you implement the syntax, like ```sql CREATE PROCEDURE proc() PATH 'test2'...

I see, thanks

@ottok nice, thanks. Because currently we don't check for libfmt version, but verify whether `{:L}` formats the number as expected, it means mariadb will use your patched libfmt just fine.

@chenrui333 , do I understand correctly, that you need this to make MariaDB build in homebrew? And "improve libfmt detection" means that without this patch cmake cannot detect system libfmt...

I generally disagree with the [assert != assume whitepaper](https://www.open-std.org/jtc1/sc22/wg21/docs/papers/2020/p2064r0.pdf), I think they use asserts incorrectly, and think that assert = assume. Still, the point on whether it will actually improve...

1. agree, if we assert an expression that uses debug-only variables, it won't compile without debug, there's nothing to assume. 2. `ut_a(!pthread_mutex_lock(&m_mutex));` — I guess, it's not a debug assert...

I've just — few days ago — added `field_flags` array to `mariadb-dump`. In my commit it only stores whether the first is VECTOR or not, but the idea was that...

Not quite, I meant this: ```diff --- a/client/mysqldump.cc +++ b/client/mysqldump.cc @@ -104,7 +104,7 @@ #define DUMP_TABLE_SEQUENCE 1 /* until MDEV-35831 is implemented, we'll have to detect VECTOR by name */...