Add support for DELETE query planning

[seonWKim]

Purpose of this PR

Support for DELETE query execution planning

Implementation Details

The main entry point for planning DELETE queries is the translate_delete function. It is composed of three primary steps:

• prepare_delete_plan:

  • Reuses the existing SELECT query's WHERE clause parsing logic to interpret and construct the initial delete plan.

• optimize_delete_plan:

  • eliminating BETWEEN expressions
  • usage of indexes

• emit_program_for_delete:

  • Add instructions for delete operation

I've tried to reuse existing logic(mostly on SELECT operations) as much as I can, so that we can automatically incorporate new changes automatically.

Delete planning debug

I've used println!(...) to specify the rows to delete. Example below

Bytecode compatibility

EXPLAIN DELETE FROM users WHERE id = 1;

EXPLAIN DELETE FROM users WHERE id > 3

EXPLAIN DELETE FROM users WHERE id < 3

TODO(future works)

• Add support for Clear opcode
• Add test when delete is implemented in Cursor

[jussisaurio]

This should be fairly simple to add, search for DecrJumpZero in emitter.rs and put one of those in emit_delete_insns()

[seonWKim]

I want to maintain this PR small for now. And I want to improve my understanding on limit(+ order by).

What do you think of I add a follow up PR right after?

[seonWKim]

Changed my mind. Added limit support 🤖

[jussisaurio]

@penberg do you have any infinite wisdom about what we should do about this PR? the query planner part for DELETE looks good, but since there is no actual delete implementation, it just prints which rowids it would have deleted. should we merge this behind a default-off feature flag to reduce confusion in case someone actually uses the lib

[penberg]

@jussisaurio @seonWKim I think we can just merge this (as soon as someone fixes the merge conflicts). We can hide it under a general-purpose feature flag like "experimental" or "preview" but I don't think that's necessarily needed right now. We're at 0.0.10 so hopefully people understand that many things just don't work.

[seonWKim]

@penberg I'll resolve the conflict and let you know

[seonWKim]

#455