Repository navigation
Question about fix timeline for DuckDB-WASM remote Parquet partial loading regression with presigned URLs #2247
Unanswered
missserious
asked this question in
Q&A
Replies: 0 comments
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Uh oh!
There was an error while loading. Please reload this page.
Title:
Remote Parquet partial loading regression in DuckDB-WASM with presigned URLs - fix timeline?
Body:
Dear DuckDB team,
I am currently using DuckDB-WASM in a web application to query remote GeoParquet files directly from object storage.
I noticed that the regression affecting remote Parquet partial loading in newer DuckDB-WASM versions is already known from existing issues.
For my use case,
@duckdb/duckdb-wasmversion1.29.0is currently the latest version that works reliably. Newer versions affect the partial loading behaviour, which is important for our application because we rely on HTTP Range Requests to efficiently query large remote Parquet files.I have also tested the latest development version, but the partial loading behaviour does not appear to be fixed there yet.
I wanted to ask:
1.29.0until the fix is released?Our setup:
1.29.0(working)pidThe same setup is used when comparing different DuckDB-WASM versions.
With
v1.29.0, the same files and queries correctly use HTTP Range Requests and only fetch the required byte ranges.Observed behaviour with newer versions:
Instead of using HTTP Range Requests to load only the required parts of the Parquet file, the complete Parquet file is loaded as a bulk download.
Has anyone else observed similar behaviour, or is there an expected timeline for a fix?
I would be happy to provide a minimal reproduction example or additional details if this helps. Also I would be happy to help with testing potential fixes or provide feedback on development versions if that would be useful.
Thanks a lot for your work on DuckDB-WASM.
Thank you so much! 💚
All reactions