embucket-labs
embucket-labs copied to clipboard
Experimental version. A BYOC option for Snowflake workloads
At the moment, the only supported credential type is access key. Extend s3 based volumes to accept either user provided role arn (and assume this role), or be able to...
snowsql --config ~/.snowflake/config.toml -c local-embucket -q "show schemas" ``` * SnowSQL * v1.4.3 Type SQL statements or !help +------------+--------------------+------+---------------+-------------+ | created_on | name | kind | database_name | schema_name |...
**Function:** `to_double` **Category:** Conversion **Status:** Failed (Return Code: 1) **Documentation:** https://docs.snowflake.com/en/sql-reference/functions/to_double **Test SQL:** ```sql SELECT TO_DOUBLE('123.45'); ``` **Expected Result:** ``` N/A ``` **Actual Result:** ``` SELECT TO_DOUBLE('123.45'); ``` **Stderr:** ```...
Getting non-clear error: 000200: 200: SQL compilation error: unsupported feature: Unsupported SQL type Character(Some(IntegerLength { length: 10, unit: None })) Expected error, as in Snowflake: 001003 (42000): 01bdbb3d-0003-340a-0004-d66e02cf2262: SQL compilation...
Embucket returns non descriptive error (I forgot to add quote for string), like this: ` SQL error: ParserError("Expected: another copy option, ; or EOF', found: EOF")` For following SQL :...
`SHOW DATABASES` shows no s3tables database, and thus it can't be used for anything. But it works upon Embucket restart, as volume already exists.
When s3table bucket has no assigned s3tables:PutTableData permission and user runs `CREATE TABLE` sql - it fails, but accidentally creates table artifact in s3table bucket. Then Embucket fails on subsequent...
Test fails ``` #[tokio::test] #[allow(clippy::too_many_lines)] #[should_panic( expected = "Failed to get error response: reqwest::Error { kind: Decode, source: Error(\"missing field `message`\", line: 1, column: 122) }" )] ``` link: https://github.com/Embucket/embucket/blob/6bcc863e380eb68a9850d00fa50c9249ea9d8883/crates/api-ui/src/tests/databases.rs#L17
Right now, the cast rule is done as a Logical `AnalyzerRule`, but it may be done as any of the following (decide which is more correct): - As is -...
We need to support the casting as `to_*` UDF calls to adhere to compatibility, not relying on DF cast. Now, this is supported via a **Logical** `AnalyzerRule` as `CastAnalyzer` (may...