This PR fixes issues on WHERE clauses for Materialized Views.
Ultimately, the problems stem from how we had a positive list of what is
a "complex expression" (needs projection), defaulting everything else to
simple (doesn't need projection). I guess it is fair to call me naive.
It should obviously be the other way. Most expressions need computation
=)
Closes#3396
We had code for this, but the code had a fatal flaw: it tried to detect
a complex operation (an operation that needs projection), and return
false (no need for projection), for the others.
This is the exact opposite of what we should do: we should identify the
*simple* operations, and then return true (needs projection) for the
rest.
CAST is a special beast, since it is not a function, but rather, a
special opcode. Everything else above is the true just the same. But for
CAST, we have to do the extra work to capture it in the logical plan and
pass it down.
Fixes#3372Fixes#3370Fixes#3369
Closes#3392 , Closes#3395
1. Prevent `ALTER TABLE t DROP COLUMN a` when exists e.g. `CREATE INDEX
ta ON t(a)`
2. Prevent `ALTER TABLE t DROP COLUMN a` when exists e.g. `CREATE INDEX
tba ON t(b) WHERE t.a > 100`
3. Prevent `ALTER TABLE t DROP COLUMN a` when exists e.g. `CREATE VIEW
ta AS SELECT a FROM t`;
and
4. Prevent `ALTER TABLE t RENAME a TO b` when exists e.g. `CREATE VIEW
ta AS SELECT a FROM t`;
Number 4 is incompatible with SQLite because we should be rewriting the
view to include the renamed column. I left that as a FIXME
Closes#3394
`CAST('yyyy-mm-dd' as datetime)` causes sqlite and tursodb to interpret
the value as just 'yyyy', e.g. '1995-01-01' becomes '1995', causing a
lot of the queries not to return any results, which is not what we want.
Reviewed-by: Preston Thorpe <preston@turso.tech>
Closes#3374
If we open database and logical log is not empty we need to recover from
it. We also make sure a single recover executes concurrently and other
connections just wait for it to finish.
I also changed the fuzz tester to use `restart` instead of calling
`load_logical_log` manually to test this behaviour.
Reviewed-by: Jussi Saurio <jussi.saurio@gmail.com>
Closes#3359
This PR contains _no semantic changes_.
I made this cleanup on another branch when I was working on subqueries,
as the inconsistency with passing around `Schema` and `SymbolTable` had
been kinda bothering me. By adding `param_ctx` to the program, it
prevents accidentally resetting it to zero.
Reviewed-by: Jussi Saurio <jussi.saurio@gmail.com>
Closes#3382
- ignore NULL rows as SQLite do
- emit NULL instead of NaN when no rows were aggregated
- adjust agg column alias name
Reviewed-by: Jussi Saurio <jussi.saurio@gmail.com>
Closes#3376
this causes sqlite and tursodb to interpret the value as just 'yyyy',
e.g. '1995-01-01' becomes '1995', causing a lot of the queries not to
return any results, which is not what we want.
