Closes#744
```sql
# Wanda = 9, Whitney = 11, William = 111
do_execsql_test column_alias_in_group_by_order_by_having {
select first_name as fn, count(1) as fn_count from users where fn in ('Wanda', 'Whitney', 'William') group by fn having fn_count > 10 order by fn_count;
} {Whitney|11
William|111}
```
Closes#864
Despite likely replacing this in-memory IO setup in the near future with
a `mmap` implementation (#859) , in the spirit of everyone getting
bitten by the perf bug lately I thought I would speed up our in-memory
IO a bit.
Closes#861
We spend a lot of time especially in `GROUP BY` queries providing
helpful comments for `EXPLAIN`, even when the query is not an `EXPLAIN`.
So let's not do that
Closes#784
```sql
Prepare `SELECT first_name, count(1) FROM users GROUP BY first_name HAVING count(1) > 1 ORDER BY cou...
time: [4.2724 µs 4.2783 µs 4.2848 µs]
change: [-6.1063% -5.7376% -5.3626%] (p = 0.00 < 0.05)
Performance has improved.
```
doesn't affect the other trivial prepare benchmarks
Closes#875
If you have a `&str` you would need to allocate and copy the string just
to pass a reference to it again. Same goes if you have a slice of bytes.
In all (most?) situations, that is not what you want and sometimes
impossible to satisfy. Example:
```rs
impl From<&str> for Value<'_> {
fn from(value: &str) -> Self {
Self::Text(&value.to_owned())
}
}
```
Here, there is no way to pass a reference to a `String` without making
the lifetime `'static`, since the string has to be dropped by the end of
the function or leaked. I would consider this a anti-pattern. There is
no reason to keep a shared reference to a owned value. (And can't think
of any situation where you would need such thing)
Now, this is possible:
```rs
impl<'a> From<&'a str> for Value<'a> {
fn from(value: &'a str) -> Self {
Self::Text(value)
}
}
```
Reviewed-by: Jussi Saurio <jussi.saurio@gmail.com>
Closes#838
Fix few logical codegen issues and add fuzz tests for logical
expressions
- Right now Limbo fails to recognize `false` constant in case when any
unary operator is used on the AST path. This PR add unary operator
option in the rewrite code and handle such cases properly.
```sql
limbo> SELECT NOT FALSE;
× Parse error: no such column: FALSE - should this be a string literal in single-quotes?
```
- `ifnull` implementation produced incorrect codegen due to "careless"
management of registers
```
limbo> SELECT ifnull(0, NOT 0)
[NULL here]
```
- `like` implementation produced incorrect codegen due to "careless"
management of registers
```
limbo> SELECT like('a%', 'a') = 1;
thread 'main' panicked at core/vdbe/mod.rs:1902:41:
internal error: entered unreachable code: Like on non-text registers
note: run with `RUST_BACKTRACE=1` environment variable to display a backtrace
```
Depends on https://github.com/tursodatabase/limbo/pull/867 (need
`GrammarGenerator` from this branch)
Closes#869
This PR fixes a bug when index search used incorrect operator if index
column were the "rhs" in the expression (not "lhs" as usual, e.g.
`SELECT * FROM t WHERE 1 < rowid_alias`)
Reviewed-by: Jussi Saurio (@jussisaurio)
Closes#870
This PR fixes implementation of binary shift right/left instructions.
Before there were a minor incompatibility between limbo and sqlite
implementation in case when right shift second argument were more than
64 and first argument were negative. As sqlite implementation of right
binary shift is sign-extended - so `-1` should be returned in such case
when limbo returned zero.
This PR fixes this bug and also introduce a fuzz tests for arithemtic
expressions. This fuzz test were written with a help of
`GrammarGenerator` which allows to easily define probabilistic context-
free grammar and then later sample random strings from it.
Closes#867
Fix panic in case when table has row with rowid equals to `-1`
(`=u64::max`)
```sql
limbo> CREATE TABLE t(x INTEGER PRIMARY KEY)
limbo> INSERT INTO t VALUES (-1)
limbo> INSERT INTO t VALUES (NULL);
thread 'main' panicked at core/vdbe/mod.rs:2499:21:
attempt to add with overflow
note: run with `RUST_BACKTRACE=1` environment variable to display a backtrace
```
Reviewed-by: Pere Diaz Bou <pere-altea@homail.com>
Closes#868
We transform all JOIN conditions into WHERE clause terms in the query
planner. The JoinAwareConditionExpr name tries to make that point, but I
think it makes things more confusing. Let's call it WhereTerm (suggested
by Jussi).
We don't need `push_predicates()` because that never REALLY was a predicate
pushdown optimization -- it just pushed WHERE clause condition expressions
into the correct SourceOperator nodes in the tree.
Now that we don't have a SourceOperator tree anymore and we keep the conditions
in the WHERE clause instead, we don't need to "push" anything anymore. Leaves
room for ACTUAL predicate pushdown optimizations later :)
We also don't need any weird bitmask stuff anymore, and perhaps we never did,
to determine where conditions should be evaluated.
Now that we do not have a tree of SourceOperators but rather
a Vec of TableReferences, we can just use loops instead of
recursion for handling the main query loop.
- use new TableReference and JoinAwareConditionExpr
- add utilities for determining at which loop depth a
WHERE condition should be evaluated, now that "operators"
do not carry condition expressions inside them anymore.
- Get rid of SourceOperator tree
- Make plan have a Vec of TableReference, and TableReference now
contains the information from the old SourceOperator.
- Remove `predicates` (conditions) from Table References -- put
everything in the WHERE clause like SQLite, and attach metadata
to the where clause expressions with JoinAwareConditionExpr struct.
- Refactor select_star() to be simpler now that we use a vec, not a tree
The main difference between = and != is how null values are handled.
SQLite passes a flag "NULLEQ" to Eq and Ne to disambiguate that.
In the presence of that flag, NULL = NULL.
Some prep work is done to make sure we can pass a flag instead of a
boolean to Eq and Ne. I looked into the bitflags crate but got a bit
scared with the list of dependencies.
Warning:
The following query produces a different result for Limbo:
```
select * from demo where value is null or id == 2;
```
I strongly suspect the issue is with the OR implementation, though. The
bytecode generated is quite different.
Reviewed-by: Jussi Saurio <jussi.saurio@gmail.com>
Closes#847
Instead of always having the caller specify all instructions, this
work introduces convenience functions into the program builder,
making the code a lot cleaner.
Draft for now, as this is done on top of #841
Reviewed-by: Jussi Saurio <jussi.saurio@gmail.com>
Closes#844
The main difference between = and != is how null values are handled.
SQLite passes a flag "NULLEQ" to Eq and Ne to disambiguate that.
In the presence of that flag, NULL = NULL.
Some prep work is done to make sure we can pass a flag instead of a
boolean to Eq and Ne. I looked into the bitflags crate but got a bit
scared with the list of dependencies.
Instead of always having the caller specify all instructions, this
work introduces convenience functions into the program builder,
making the code a lot cleaner.
The error is due to comparing the PRIMARY KEY's name to INTEGER when in
it was all in lowercase. This was causing `needs_auto_index` to be set
to `true`.
After the fix:
```
/limbo /tmp/sc2-limbo.db
Limbo v0.0.13
Enter ".help" for usage hints.
limbo> CREATE TABLE temp (t1 integer, primary key (t1));
hexdump -s 28 -n 4 /tmp/sc2-limbo.db
000001c 0000 0200 -- matches SQLite
0000020
```
Closes https://github.com/tursodatabase/limbo/issues/824
Reviewed-by: Jussi Saurio <jussi.saurio@gmail.com>
Closes#830
The pragma list will only grow. The strum crate can be used to:
* automatically convert to string from enum
* automatically convert to enum from string
* implement an iterator over all elements of the enum
