Pekka Enberg

We need time querying and random number generation to be deterministic for simulation. Therefore, move them to the I/O trait so that we can simulate them.

In-memory mode

5

Add support for in-memory mode like SQLite has because it's great for just playing around with a database.

Port SQLite TCL tests

2

The SQLite source tree has bunch of TCL tests that we ought to port to run with our TCL runner.

Date and time function support

3

This is a tracking issue for all date and time functions: https://github.com/penberg/limbo/blob/main/COMPAT.md#date-and-time-functions

Scalar function support

3

This is a tracking issue for all scalar functions: https://github.com/penberg/limbo/blob/main/COMPAT.md#scalar-functions

We currently use the `readSync()` and `writeSync()` Node filesystem APIs. However, Limbo's I/O completion model should allow us to use the callback-based APIs, which don't block the Node I/O loop.

External sorting for `ORDER BY`

10

All the sorting now happens in-memory, which won't work well for large data sets. Let's add support for external sorting too.

Rowid is not supported in WHERE clause

2

``` sqlite> EXPLAIN SELECT rowid FROM users WHERE rowid = 1; addr opcode p1 p2 p3 p4 p5 comment ---- ------------- ---- ---- ---- ------------- -- ------------- 0 Init 0...

As discussed in this issue, some users may want to consume Limbo on Python via a pylimbo package: https://github.com/penberg/limbo/issues/128#issuecomment-2255253421 We can build a `bindings/python` package using https://www.maturin.rs, which follows the...

CPython support

3

This is going to be a huge task, but the goal is to be able to build CPython using Limbo's SQLite compatibility library. You can build Python with Limbo as...