This PR is based on #1357 and further improves performance:
```sql
limbo> select l_orderkey, 3 as revenue, o_orderdate, o_shippriority from lineitem, orders, customer where c_mktsegment = 'FURNITURE' and c_custkey = o_custkey and l_orderkey = o_orderkey and o_orderdate < cast('1995-03-29' as datetime) and l_shipdate > cast('1995-03-29' as datetime);
┌────────────┬─────────┬─────────────┬────────────────┐
│ l_orderkey │ revenue │ o_orderdate │ o_shippriority │
├────────────┼─────────┼─────────────┼────────────────┤
└────────────┴─────────┴─────────────┴────────────────┘
Command stats:
----------------------------
total: 3.728050958 s (this includes parsing/coloring of cli app)
```
Reviewed-by: Preston Thorpe (@PThorpe92)
Reviewed-by: Pere Diaz Bou <pere-altea@homail.com>
Closes#1358
In left joins, even if the join condition is not matched, the system
must emit a row for every row of the outer table:
-- this must return t1.count() rows, with NULLs for all columns of t2
SELECT * FROM t1 LEFT JOIN t2 ON FALSE;
Our logic for clearing the null flag was to do it in Next/Prev. However,
this is problematic for a few reasons:
- If the inner table of the left join is using SeekRowid, then Next/Prev
is never called on its cursor, so the null flag doesn't get cleared.
- If the inner table of the left join is using a non-covering index seek,
i.e. it iterates its rows using an index, but seeks to the main table
to fetch data, then Next/Prev is never called on the main table, and the
main table's null flag doesn't get cleared.
What this results in is NULL values incorrectly being emitted for the
inner table after the first correct NULL row, since the null flag is
correctly set to true, but never cleared.
This PR fixes the issue by clearing the null flag whenever seek() is
invoked on the cursor. Hence, the null flag is now cleared on:
- next()
- prev()
- seek()
closes#1185
## The Problem:
The underlying schema of virtual tables is hidden from the query
planner, and it currently has no way of optimizing select queries with
vtab table refs by using indexes or removing non-constant predicates.
All vtabs are currently rewound completely each time and any conditional
filtering is done in the vdbe layer instead of in the `VFilter`.
## The solution:
Add xBestIndex to the vtab module API to let extensions return some
`IndexInfo` that will allow the query planner to make better
optimizations and possibly omit conditionals
## Examples:
table `t`: vtab: (key, value)
table `t2`: table: (a,b)
### Join where vtab is outer table:
Properly pushes predicate to VFilter, which receives the idx_str
`key_eq` arg, telling it that there is a useable where clause on the key
"index"
### Join where vtab is inner table:
Constraint is not sent because it is marked as unusable
### Where clause on "indexed" column:
Pushed down and the predicate is omitted from the VDBE layer.
### Where clause on regular column:
No idx info received from BestIndex, VDBE handles conditional.
## TODO:
OrderBy info needs to be sent to xBestIndex and its not in a great
position in `open_loop` currently
Reviewed-by: Jussi Saurio <jussi.saurio@gmail.com>
Closes#1264
Apply affinities to a range of P2 registers starting with P1.
P4 is a string that is P2 characters long. The N-th character of the string indicates the column affinity that should be used for the N-th memory cell in the range.
If P4==0 then register P3 holds a blob constructed by MakeRecord. If P4>0 then register P3 is the first of P4 registers that form an unpacked record.
Cursor P1 is on an index btree. If the record identified by P3 and P4 is not the prefix of any entry in P1 then a jump is made to P2. If P1 does contain an entry whose prefix matches the P3/P4 record then control falls through to the next instruction and P1 is left pointing at the matching entry.
This operation leaves the cursor in a state where it cannot be advanced in either direction. In other words, the Next and Prev opcodes do not work after this operation.
Sqlite reference implementation: https://github.com/sqlite/sqlite/blob/8
37dc09bce7de8971c7488b70cf5da93c60fbed0/src/vdbe.c#L2558
We did not support blobs before and our ordering of the match statements
were incorrect when one of the arguments was NULL
Closes#1337
Fixes#1298
- Fixes Limbo trying to use an index using a WHERE clause constraint
that refers to the same table on both sides, e.g. `WHERE t.x = t.x`
- Fixes not using indexes when the relevant expression is paren wrapped,
e.g.
- `SELECT * FROM t WHERE (indexcol) > 5`
- `SELECT * FROM t WHERE (indexcol > 5)`
- Changes existing table logical expr fuzz test to have primary keys
(which helped me find both issues above)
Closes#1300
