Skip to content

fix: restore write encoding perf by encoding vectors before transform walk #265

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Closed
rkruze wants to merge 1 commit into from
Closed

fix: restore write encoding perf by encoding vectors before transform walk #265

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
14 changes: 8 additions & 6 deletions src/turbopuffer/_utils/_transform.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -182,10 +182,11 @@ class Params(TypedDict, total=False):
It should be noted that the transformations that this function does are not represented in the type system.
"""
# turbopuffer: Use simple vector encoding instead of generic type-based transform.
data = cast(_T, _turbopuffer_transform(data))
# Encode vectors before walking the body so the walk sees one base64 string
# per vector instead of recursing into every float.
if expected_type is NamespaceWriteParams:
return cast(_T, _encode_write_vectors(data))
return data
data = cast(_T, _encode_write_vectors(data))
return cast(_T, _turbopuffer_transform(data))


@lru_cache(maxsize=8096)
Expand Down Expand Up @@ -404,10 +405,11 @@ class Params(TypedDict, total=False):
It should be noted that the transformations that this function does are not represented in the type system.
"""
# turbopuffer: Use simple vector encoding instead of generic type-based transform.
data = cast(_T, _turbopuffer_transform(data))
# Encode vectors before walking the body so the walk sees one base64 string
# per vector instead of recursing into every float.
if expected_type is NamespaceWriteParams:
return cast(_T, _encode_write_vectors(data))
return data
data = cast(_T, _encode_write_vectors(data))
return cast(_T, _turbopuffer_transform(data))


async def _async_transform_recursive(
Expand Down
Loading