This PR implements more sophisticated algorithm in the toy vector sparse
index: now we enumerate components based on the frequency (in order to
check unpopular "features" first) and also estimate length threshold
which can give us better results compared with current top-k set.
Also, this PR adds optional `delta` parameter which can enable
approximate search which will return results with score not more than
`delta` away from the optimal.
In order to implement this index method - index code were slightly
adjusted in order to allow to store some non-key payload in the index
rows. So, now index can hold N columns where first K <= N columns will
be used as identity (before that K always was equal to N).
Reviewed-by: Jussi Saurio <jussi.saurio@gmail.com>
Closes#3862
thread 'query_processing::test_read_path::test_stmt_reset' panicked at core/storage/sqlite3_ondisk.rs:754:9:
assertion failed: self.page_type() == PageType::TableLeaf
note: run with `RUST_BACKTRACE=1` environment variable to display a backtrace
test query_processing::test_read_path::test_stmt_reset ... FAILED
The SQLite varint specification states that the varint is guaranteed to be a maximum of 9 bytes, but our version of write_varint initializes a buffer of 10 bytes. Changing the size to match the specification.
This separates fuzz tests from integration tests so that you can run the
fast test cases with:
```
cargo test --test integration_tests
```
and the longer fuzz cases with:
```
cargo test --test fuzz_tests
```
The collation fuzz test case takes up to 4 minutes to run, making it the
slowest of all the test cases. Let's reduce iteration count a bit to
make this more CI friendly.
DEFERRED was a bit too deferred - it allowed the dirty pages to be
written out to WAL before checking for violations, resulting in the
violations effectively being committed even though the transaction ended
up aborting
Closes#3784Closes#3785
Rolling back a transaction should result in `connection.auto_commit` being set
back to true.
Added a regression test for this where a UNIQUE constraint violation rolls back
the transaction and trying to COMMIT will fail.
Currently, our default conflict resolution strategy is ROLLBACK,
which ends the transaction. In SQLite, the default is ABORT, which rolls back
the current statement but allows the transaction to continue.
We should migrate to default ABORT once we support subtransactions.
op_destroy was assuming we never yield IO from BTreeCursor::btree_destroy(),
so every so often it would just not complete the procedure and leave
dangling pages in the database
This PR add proper program abort in case of unfinished statement reset
and interruption.
Also, this PR makes rollback methods non-failing because otherwise of
their callers usually unclear (if rollback failed - what is the state of
statement/connection/transaction?)
Reviewed-by: Preston Thorpe <preston@turso.tech>
Closes#3591
