V1.4.3 merge #212
Merged
V1.4.3 merge #212
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This PR bumps the following extensions: - `delta` from `0747c23791` to `6515bb2560` - `ducklake` from `022cfb1373` to `77f2512a67` - `httpfs` from `b80c680f86` to `041a782b0b`
…lection operations (duckdb#19718) Fixes duckdb#19700 This probably should be maintained during the actual select - but for now just sorting it afterwards solves the issue.
…b#19713) The `PersistentColumnData` constructor asserts that the pointers aren't empty. This assertion will fail if we try to serialize the child of a list, if all lists are empty (as the child will be entirely empty then) Backported fix for problem found by: duckdb#19674
This allow explicit re-initialization of specific parts of HTTPClient(s)
This diff would allow patterns such reusing partially constructed (but
properly re-initialized) HTTPClient objects
```c++
struct CrossQueryState {
// in some state kept around
unique_ptr<HTTPClient>& client;
};
void SomeFunction() {
// ...
http_util.Request(get_request, client);
// some more logic, same query
http_util.Request(get_request, client);
}
void SomeOtherFunction() {
// Re-initialize part of the client, given some settings might have changed
auto http_params = HTTPParams(http_util)
client->Initialize(http_params);
// ...
http_util.Request(get_request, client);
// some more logic, same query
http_util.Request(get_request, client);
}
```
Note that PR is fully opt-in from users, while if you implement a
file-system abstraction inheriting from HTTPClient you should get a
compiler error pointing to implementing the relevant function.
…9734) Currently `CALL enable_logging('http');` would succeed, but then select an empty subset of the available logs (`http` != `HTTP`), due to a quirk in the code. This PR fixes that up.
…that the child must have columns, which doenst have to be true if the expression is a constant
Closes duckdb#19645 The issue is that during pipeline verification we compare all the operators across pipelines and ensure that their equality is reflexive (which is kind of a strange check actually), but in the case of table scans we only compare the normal table function callback pointer. In the MRE in the linked issue, the pipeline contains a UNNEST and a RANGE, which are both table-in-out functions, but not _normal_ table functions, so the function pointer(s) are null on both sides and they are assumed to be equal. I've fixed it by actually implementing the `==` and `!=` operates for `TableFunction`s that compare all the properties, similar to how its done for `ScalarFunction`.
This PR supersedes duckdb#19625 and fixes duckdb#19313 We now create a LogicalAggregate even if we only have a HAVING clause, no aggregates, no group by. This matches PostgreSQL behavior.
… side have correlation Kind of a weird case, maybe we should leave the join decorrelation early here. Will have to investigate some time.
Fixes nightly failure found here https://github.com/duckdb/duckdb/actions/runs/19842531250/job/56857957914#step:6:4105
* Reverse the steaming/blocking window operator generation order.
…its::Min so that stats are initialized to -infinity for floating points (duckdb#20039) Fix duckdb#20014
This change adds `opt_in_archs` argument to the `Extensions.yml` workflow and passes it down to `_extension_distribution.yml`. The default value is set to `windows_arm64;` with the intention to keep it there (if it will work correctly and will not cause problems) to have Windows ARM64 extension builds enabled by default. This change only enables it for Main extensions. Rust-based extensions (`delta`) can be added later - it requires Python wheel re-build (for `configure` step, until `windows_arm64` binary wheel is available). It works on CI, but takes additional 35 mins.
Mark joins without correlation on the left side could not be handled correctly without this PR. Information regarding the correlated columns was simply lost, which could lead to wrong query behavior. This PR fixes this and introduces correct decorrelation of mark joins without correlation on the left side. ---- Fix: duckdb#19504
Bugfix for duckdb#19975 Propagate gate status in Node4::DeleteChild, since Prefix::Concat logic depends on both the prev_node4 gate status (was the node4 being freed a gate node), as well as the overall gate status (did we enter through a gate at some point earlier). The latter was not being propagated correctly, and node compression logic for nested ARTs storing rowids depends on this, since we do not need to maintain prefix chains, as they are implicit in the rowid's. Unfortunately it is not possible to create a minimal test for this scenario, however once duckdblabs/duckdb-internal#6083 is solved, there is a large reproducer for this scenario that I downloaded from the person who opened this issue (and the data/sql file is not confidential, so can be used). Thank you @taniabogatsch for dealing the final blow to this bug!
This PR bumps the following extensions: - `httpfs` from `8ff2283fb1` to `9c7d34977b`
During CTE decorrelation, the `RewriteCTEScan` class adds correlated columns (in case of multiple dependent joins) to child dependent joins. However, this is only required for recursive CTEs. ---- Fix: duckdb#20016
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
20 participants
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
No description provided.