Thijs
Thijs
I could see it making sense to create a basic implementation that does a bind on the sql query as-is, if that fails we stop Then check if the produced...
Our current planning does not include any work on this feature, no clue when that will change
Debatable whether this is a bug, I figured since we support named in-memory connections, ATTACH should have the same behavior. So I would consider this behavior unexpected
Are you sure you're not pinging the wrong OAuth2 server? There's an ATTACH option to change the OAuth2 server, by default it uses a path provided by the REST spec,...
My hunch is that it's not the avro files, but the `metadata.json` that is "too large", which would make sense given that you mentioned the table has 7319 snapshots
That honestly sounds like a bug in duckdb-jdbc If we look at we can see this error is thrown when the GetTransaction() method is called on a DuckLake scan after...
Or use a generic protocol, perhaps ODBC which is supported through a community extension (https://duckdb.org/community_extensions/extensions/nanodbc.html) ?
This was the old behavior, before TableFilterType::EXPRESSION existed, I'm not sure why this got lost when EXPRESSION was introduced
Hmm that is odd, I imagine the stats get pushed down incorrectly, so it tries to evaluate the filters through stats and determines the result to be always false
Note to self: we can apply this same process to resolve the merge conflicts with . In the same way we push the struct extract down to the storage, we...