Skip to content

Commit

Permalink
Rewrite CloudWriter to use BufWriter from object_store
Browse files Browse the repository at this point in the history
This makes the "ObjectStore" choose to upload using a single request or
a multi request (multi part) upload depending on the size of the chunks.

This change is needed because Polars is now calling the writer without
buffering, so this was breaking the upload.

There are two tests failing for different reasons:
- the IPC upload to a unknown bucket is failing because the new
  CloudWriter is not propagating the error. This is probably an easy
  fix.
- the Lazy "parquet to cloud" is failing for the reason I wrote above.
  This is probably related to this issue:

  pola-rs/polars#17172
  • Loading branch information
philss committed Jul 20, 2024
1 parent 878b422 commit 0b1f39e
Show file tree
Hide file tree
Showing 6 changed files with 66 additions and 237 deletions.
166 changes: 26 additions & 140 deletions native/explorer/Cargo.lock

Some generated files are not rendered by default. Learn more about how customized files appear on GitHub.

3 changes: 1 addition & 2 deletions native/explorer/Cargo.toml
Original file line number Diff line number Diff line change
Expand Up @@ -30,7 +30,7 @@ tokio-util = { version = "0.7", default-features = false, features = [
"io",
"io-util",
], optional = true }
object_store = { version = "0.9", default-features = false, optional = true }
object_store = { version = "0.10", default-features = false, optional = true }

# MiMalloc won´t compile on Windows with the GCC compiler.
# On Linux with Musl it won´t load correctly.
Expand All @@ -55,7 +55,6 @@ features = [
"ewma",
"extract_groups",
"extract_jsonpath",
# "group_by_list",
"ipc_streaming",
"ipc",
"is_in",
Expand Down
Loading

0 comments on commit 0b1f39e

Please sign in to comment.