I still didn't find a good way to implement variadic functions, we
should have some sort of wrapper in JS layer but it didn't work so well
for me so far. But once done it will be easily transferable to any
function.
It also should probably be async, but AFAIC napi doesn't have a straight
way to implement async iterators.
Closes#1515
Various things to improve speed of long fuzz test execution time:
* remove unnecessary debug_validate_cell calls
* Add SortedVec for keys in fuzz tests
* Validate btree's depth in fuzz test every 1K inserts to not overload
test with validations. We add `VALIDATE_BTREE` env variable to enable
validation on every insert in case it is needed.
Reviewed-by: Jussi Saurio <jussi.saurio@gmail.com>
Closes#1521
1. `group_by_contains_all` was incorrect - it was not checking that all
order by columns are in group by; it was instead checking that all group
by columns are in order by, which is absolutely incorrect for the
intended purpose.
2. remove ORDER BY clause if GROUP BY clause can sort the rows in the
same way.
Test failures are not related
Reviewed-by: Pere Diaz Bou <pere-altea@homail.com>
Closes#1511
We've run into trouble in multiple places due to the fact that we delete
terms from the where clause (e.g. when a constant condition is removed,
or the term becomes part of an index seek key).
A simpler solution is to add a flag indicating that the term is consumed
(used), so that it is not translated in the main loop anymore when WHERE
clause terms are evaluated.
note: CI failures are unrelated
Reviewed-by: Pere Diaz Bou <pere-altea@homail.com>
Closes#1477
Various things:
* remove unnecessary debug_validate_cell calls
* Add SortedVec for keys in fuzz tests
* Validate btree's depth in fuzz test every 1K inserts to not overload
test with validations. We add `VALIDATE_BTREE` env variable to enable
validation on every insert in case it is needed.
This PR adds a new function `read_write_payload_with_offset` to support
reading and writing payload data at specific offsets, handling both
local content and overflow pages. This is a port of SQLite's
`accessPayload` function in `btree.c` and will be essential for
supporting incremental blob I/O in the coming PRs.
- Added a state machine called `PayloadOverflowWithOffset` to make the
procedure reentrant.
- Correctly processes both local payload data and payload stored in
overflow pages
Testing:
- Reading and writing to a column with no overflow pages.
- Reading and writing at an offset with overflow pages (spanning 10
pages)
Reviewed-by: Pere Diaz Bou <pere-altea@homail.com>
Closes#1476
As there were many merge conflicts for the other PR, I rewrote the code
and condensed it here.
ORIGINAL PR TEXT: Provides the code to almost close
https://github.com/tursodatabase/limbo/issues/1251 . The JsonSchema is
derived, but I am still not sure how to automate the distribution to
SchemaStore for autocomplete. I added some docs for that want to see the
config file description. I still am not sure how to automate this
documentation. Maybe some macro magic?
Reviewed-by: Preston Thorpe (@PThorpe92)
Closes#1430
Closes#1482. I needed to change the `key_exists_in_index` function
because it zips the values from the records it is comparing, but if one
of the records is empty or not of the same length, the `all` function
could return true incorrectly.
Reviewed-by: Jussi Saurio <jussi.saurio@gmail.com>
Closes#1514
Closes#1508 . There were two small issues to fix:
1. We were not checking in the IndexMap of columns, if the unique column
name is declared in the composite declaration exists in the IndexMap.
This solved the first this statement `create table t4(a, unique(b));`.
2. The second thing was that we forgot to add the column_name to the
HashSet of columns.
```rust
Some(PrimaryKeyDefinitionType::Simple { column, .. }) => {
let mut columns = HashSet::new();
columns.insert(std::mem::take(column));
// Have to also insert the current column_name we are iterating over in primary_key_column_results
columns.insert(column_name.clone()); <-- Fix here
primary_key_definition =
Some(PrimaryKeyDefinitionType::Composite { columns });
}
```
The rest of the modifications are just some small simplifications for
readability and avoiding some clones
Reviewed-by: Jussi Saurio <jussi.saurio@gmail.com>
Closes#1512
This prevents the new wal checkpoint tests in `sqlite3/tests/compat`
from writing/creating `test` table to `testing/testing.db`, which is
queried in later tests which fail for having an extra table.
There is another issue with failing tests related to the new `count`
impl that I am in the process of fixing as well, but that will be a
separate PR.
Closes#1513
We've run into trouble in multiple places due to the fact that
we delete terms from the where clause (e.g. when a constant condition
is removed, or the term becomes part of an index seek key).
A simpler solution is to add a flag indicating that the term is
consumed (used), so that it is not translated in the main loop
anymore when WHERE clause terms are evaluated.
I saw that the labeler workflow trying to get `labeler.yml` file but i
see it's `.yaml` in the repo so i renamed it.
Reviewed-by: Jussi Saurio <jussi.saurio@gmail.com>
Closes#1506
Primary keys that are marked as unique constraint, do not need to have
separate indexes, one is enough. In the case primary key is integer,
therefore rowid alias, we still need an index to satisfy unique
constraint.
Closes#1494
No semantic changes in this PR, the macros were organized because
`lib.rs` was getting out of hand anyhow, and because it's now easier to
determine if a change submitted in a PR is related to `extensions/core`
for labeling.
# NOTE:
this will not work until it is merged, because it needs write
permissions from the base or root of the repo
https://github.com/actions/labeler?tab=readme-ov-file#initial-set-up-of-
the-labeler-action
Closes#1503
