vtgate: Handle SET for sql_auto_is_null

[vmg]

Hi @sougou! Here's a small patch to improve compatibility with Ruby on Rails for the MySQL layer. I know you discussed this briefly with @arthurnn in chat.

I would also love it if there was a way for Vitess (probably the vttablets) to ensure that sql_auto_is_null has not been enabled manually by the user (i.e. in the MySQL configs), because the feature is insane and it's gonna break stuff. I haven't found the right place to add the check tho.

---

Many ORMs (most notably Rails' ActiveRecord and Django) start their
MySQL connections with a SET sql_auto_is_null=0.

This setting is superfluous in all MySQL engines that are supported by
Vitess, where sql_auto_is_null is always disabled by default.
Furthermore, it doesn't seem feasible to actually implement this feature
in Vitess because of sharded queries.

Because of this, we propose handling the case where the feature is set
to 0 (i.e. a no-op), and returning an explicit error when somebody tries
to actually enable the feature.

This improves the compatibility of Vitess' MySQL layer with these ORMs,
and improves the UX for users attempting to enable sql_auto_is_null in
a connection.

[sougou]

Wow. Nasty feature. We should disable it. To make vttablet enforce this, you can add the check in the VerifyMode function: https://github.com/youtube/vitess/blob/master/go/vt/vttablet/tabletserver/connpool/dbconn.go#L225.

PS: The test has a minor compilation error.

[vmg]

Ah yes, that's the perfect spot. Updated the PR!

[sougou]

Some new tests are failing. Looks like you need to add that query to fakesqldb.

[vmg]

phew! Took me a while to get the fakesqldb queries wired up everywhere. Finally got it right.

Ready to review!

[sougou]

LGTM
Yeah. Those are a pain. But they pay off because we don't have to bring up real mysql servers for them. We should look at unifying that framework to make maintenance easier.