`DeletePredicate` is a simpler version of `Predicate` that is based on
IOx `DeleteExpr` instead of the full-blown DataFusion `Expr`. This will
allow us to do a couple of things (in follow-up changes):
- Order and de-duplicate delete predicates
- Normalize predicates
- Infallible serialization
- Smaller memory footprint
Note that this change only affects delete expressions. Query expressions
that are supported via the API are not changed. The query subsystem also
still uses the full-featured expressions/predicates (delete
expressions/predicates are converted to the more powerful DataFusion
version on-the-fly).
Closes: #1603
If a predicate cannot be executed against a read buffer chunk because of schema conflicts then fall back to applying no predicate and let the query engine apply predicates in the Filter step of the plan.
* fix: Await on a freeze handle instead of returning TransitionInProgress
* fix: Await on freeze handle in skip_replay
Co-authored-by: kodiakhq[bot] <49736102+kodiakhq[bot]@users.noreply.github.com>
* chore: Update datafusion + other deps
* refactor: update query crate for new async interfaces
* refactor: update server crate for new async interface
* refactor: update query_tests crate for new async interfaces
* refactor: update ioxd and server to use new async interface
Co-authored-by: kodiakhq[bot] <49736102+kodiakhq[bot]@users.noreply.github.com>
Due to the timing of the "persist" lifecycle action and that delete
predicates might arrive at any time + the fact that we don't wanna hold
transaction locks for too long, we should accept delete predicates for
chunks that are currently "persisting" even though that lifecycle action
might fail.
The old construct uses a single assert-statement for both:
- "bubble-up" scenario, were a panic should fire
- a check, were a panic should not fire
That makes it easy to add new tests. However we need two rather
questionable things to make that work:
- catch panic: to convert an assertion to a check
- a custom panic hook: to make tests not overly verbose (aka caught
panics should not show up on stdout)
Esp. the custom panic hook doesn't work too well w/ multi-threaded tests
since it might swallow error messages from unrelated tests and makes
debugging of CI failures hard.
So instead of using assertions for checks, we now implement a proper
assertion and a check for each test. That's a bit more code per check
but easier probably more stable.
First step towards #2518. Creates the Rust API to communicate delete
predicates between the preserved catalog and the in-memory catalog and
adds tests ensuring that the in-mem catalog produces the wanted errors
as well as correct checkpoints (similar to how this is done for the
parquet file tracking already).
**This does NOT contain the actual preservation!**
kodiakhq[bot]
Raphael Taylor-Davies
Marco Neumann
Edd Robinson
dependabot[bot]
Nga Tran
Carol (Nichols || Goulding)
Carol (Nichols || Goulding)
Andrew Lamb