MariaDB character set and collation recommendations

[Wessel, Keith]

Hi, all,

In migrating to the new JDBC plugin for the IdP, I discovered that, contrary to warnings in the Shibboleth docs, our Shibboleth storage records table uses a case-insensitive character collation. This got me wondering, is there a good general purpose character set and collation to use for the storage records table? Our database, which has been around for a few years and came from a much older MySQL version before MariaDB existed, is using latin1 with its default latin1_sweedish_ci collation. Surely, some of the utf8 collations are more efficient and well-suited than latin1. Since I have to fix the case-insensitive issue, anyway, I might as well choose a better character set.

My development instance is using utf8mb4, and I just changed its collation from utf8mb4_general_ci to utf8mb4_bin. This seems like a reasonable choice, but I'd like to ask for the widom of others on the list who might be willing to share.

Thanks,
Keith