We're using a software that supports MySQL via its C connector library, version 5.7.25.

We are using MariaDB version 10.11.6 but saw the problem also on 10.11.5.

We have a table with several columns of type utf8mb4. The software retrieves column information using the C connector function mysql_fetch_field. For the utf8mb4 columns it returns MYSQL_FIELD.charsetnr == 33. This is apparently supposed to be the collation id, which they then lookup in INFORMATION_SCHEMA.COLLATIONS.

The problem is that 33 is the id for collation utf8mb3_general_ci, i.e. charset utf8mb3, which is incorrect. Correct value would, in their tests with ci collation, be id 45, collation utf8mb4_general_ci.

I'm afraid I know too little to be able to determine if this is even supposed to work, but I was under the impression that MariaDB is intended to be a fully compatible replacement to MySQL at the API level. I also know too little to be able to determine exactly where the problem lies (which component), but inspection the information in the information_schema tables, everything looks correct there, so I assume there's some problem at the api level.