Add support for DEFAULT (expression) in column definitions - #306

@wojtekn Thank you for looking into this! Good job working with the AST and even fixing the old driver 👍

Generally, it looks good to me, and it is the right approach to work with the new driver. On the AST side, I spotted one issue that is related to only implementing a subset of the expression functionality, and that is that for expressions such as this one:

DEFAULT (DATE_ADD(NOW(), INTERVAL 1 YEAR))

It would detect and incorrectly return the value of NOW(). That's because the implementation checks whether there is that function call, but it can't easily determine that there is nothing else. Fixing that is not straightforward with arbitrary nesting, e.g. for DEFAULT (((now()))), and doing the full expression support is easy, so I think we could even go all the way. Let me see if I can push my idea somehow. I probably can't modify this branch as it's from a fork, but I can push another one.

Additionally, in the MySQL column metadata, there needs to be an indication that it has a DEFAULT_GENERATED value, but I can see that we already seem to be doing that, so I think we're safe on this one. Perhaps it needs a test in case there isn't one yet.

@wojtekn It turns out I can push to this branch. I just have no idea how to manage PRs from forks 😆

I added one commit with a quick implementation of the full DEFAULT (expression) on the AST side. I will check it again after the weekend and see if it needs some polishing and/or more tests.

TODO:

• Check if we need to test more expressions.
• Add a test for metadata (DEFAULT_GENERATED ).
• Check if we can improve stored expression formatting.

Thanks again for pushing this forward!

No more feedback here, so let's get it in.

@JanJakes, thanks for taking it over and finishing it.