Development is mostly done in https://github.com/linuxJedi/mydumper/tree/mariadb-sequences

Current issue is with myloader, SEQUENCEs need to be restored before tables. This is not yet expressible in myloader which tries to restore everything in parallel. I have an idea on how to resolve this which I am passing by David Ducos.

Second issue to contend with, a relative reference to a SEQUENCE in `nextval` turns into an absolute one including the schema name. This means that backup/restore into a different database breaks.

Discussions ongoing on how to solve these two issues.

`nextval` issue is solved using a regex filter.

David wants some retry code in there so tables that fail due to missing sequence are retried after the initial table threads have run.

Some more goundwork has been laid and merged in. The MariaDB detection in mydumper now works correctly and the table detection queries for MariaDB have been changed so that sequences and system version tables can be detected correctly.

David has done the final parts of this, it is currently in review.

This has been merged upstream.