Micah Wylde

Validate filesystem URLs at planning time

3

For filesystem sources created via SQL, we do not validate them as part of the SQL planning process. This causes panics at runtime when the source is instantiated on the...

This query causes a failure at runtime ```sql CREATE TABLE mastodon ( value TEXT ) WITH ( connector = 'sse', format = 'raw_string', endpoint = 'http://mastodon.arroyo.dev/api/v1/streaming/public', events = 'update' );...

Support AWS IMDSv2

1

IMDSv2 (instance metadata service) is a new, more secure mechanism to get instance metadata from an EC2 instance: https://docs.aws.amazon.com/AWSEC2/latest/UserGuide/instance-metadata-v2-how-it-works.html. Unlike IMDSv1 which is just an endpoint that returns metadata from...

Currently the preview experience works like this: * We plan the query * We compile it into a dataflow * We traverse the dataflow and find all of the sinks...

We have two syntaxes for creating an interval: * `INTERVAL '5' second` * `'5 seconds'` Both work as arguments to window functions: ``` tumble(interval '5' second) tumble('5 seconds') ``` But...

Our SQL cast expressions currently deal with errors by unwrapping: https://github.com/ArroyoSystems/arroyo/blob/39c1bd6779f395688cc56c2fc608b9cee88170c7/arroyo-sql/src/expressions.rs#L1243 Causing an error on invalid cast is inline with the behavior of postgres, however we should find a way...

This query fails in planning: ```sql create table input ( length JSON, diff INT GENERATED ALWAYS AS (extract_json(length, '$.old')[1]) STORED ) with ( connector = 'sse', endpoint = 'https://localhost:9091', format...

Adds additional json functions, from https://github.com/datafusion-contrib/datafusion-functions-json: * json_contains(json: str, *keys: str | int) -> bool * json_get(json: str, *keys: str | int) -> JsonUnion * json_get_str(json: str, *keys: str |...

Building without heavyweight connectors shaves off ~500 transitive deps, speeding up development build cycles