Connect to database using SET NAMES utf8mb4

[mattwire]

Overview

Alternative to #17696 as discussed by @pfigel @mfb @demeritcowboy @eileenmcnaughton @seamuslee001 . The use of SET NAMES utf8mb4 has no impact on supported mysql versions for CiviCRM even if the site is not using utf8mb4.

I've switched all uses of SET NAMES (that I could find) to utf8mb4

Before

Database connection with SET NAMES utf8 - can't use utf8mb4 characters.

After

Database connection with SET NAMES utf8mb4 - can use utf8mb4 characters.

Technical Details

Comments

[civibot]

(Standard links)

• If this is your first pull-request for CiviCRM, please browse CONTRIBUTING.md for information about the development and testing processes.
• If you are reviewing this pull-request, you may wish to consult the test sites and the Review Standards (long template, short template).

[seamuslee001]

Test failure is unrelated. I think this has the broad support based on @demeritcowboy 's comment here #17696 (comment) I'm going to merge this

[demeritcowboy]

Ha ha @seamuslee001 is confusing me with @pfigel which is a win for me at least (no comment on how that will ultimately work out for you @pfigel). I suppose I'm obligated now to add something more useful to the conversation:

Joins should be fine since that wouldn't depend on SET NAMES and would depend on the COLLATION of the column definitions. So at worst it would still use the index but would slow down since it would have to convert to compare, but that's the same as regular utf8 if you define columns with different utf8_XXX collations.

Yes that sounded intelligent enough, so it doesn't even matter if it's true. (It is though - quick test confirms it - EXPLAIN shows it using index but also has the word "func", which of course means conversion).

[eileenmcnaughton]

@demeritcowboy yes high praise - but I think you deserve it :-).