Sam Ansmink

Hi @cookiejest, we currently don't support having different sets of credentials at the same time for different files. To fetch data with different credentials you will need to switch the...

@cookiejest yes this does do a copy. You can start duckdb in two ways: as a transient in-memory database or as a persistent database by passing a path: `duckdb your_db_file.db`....

@Alex-Monahan the AVX512 functions are just for speeding up the compression. Although I left the code in on purpose to be able to experiment with it a bit, I think...

@hannes Unfortunately I think that even though CI is succeeding, there is still an issue in this PR. I can reproduce this on my machine by building with `make relassert`,...

@Mytherin this PR is good to go from my side!

regards point 2: `if I don't use count(1), but instead of count(id), the runtime increases dramatically`, this is because the count(id) will actually do a scan of the id column...

will close this issue as the parquet extension is now built for macos allowing install/load

@pvaezi that would be much appreciated of course! the documentation lives here https://github.com/duckdb/duckdb-web feel free to open a PR there. There are contributing and building instructions in READMEs to help...

Hey @cpcloud thanks for reporting! This is indeed a bug. Without a secret we currently do not set the gcs endpoint correctly. A workaround is to create an empty GCS...

@adenosinew What image do you launch you instance with? For Python on ARM linux we only have extension binaries with a relatively recent glibc version (GLIBC >= 2.28). This means...