NVIDIA / NVIDIA/cudf

[FEA] Replace usage of `compile_internal` in lowering with `builder.call`

Open
#18,926 0 comments 0 reactions 0 assignees View on GitHub
feature request numba Python
Dominant language
C++
Stars
9.8k
Forks
1.1k
Avg merge
3d 6m
Merged PRs (30d)
278

Description

When implementing lowering for operations over extension types in cuDF, we make extensive use of `compile_internal` as way of arriving at the correct function call instructions that we want. The lower level way of doing this is to use `builder.call` directly. This would also decrease the complexity of our lowering and make it a bit more readable.

This may be harder in situations where we're abusing compile internal to [patch in python lambdas](https://github.com/rapidsai/cudf/blob/4655b4bcfdae9adb6d9ac045e4d95d1c74fc8888/python/cudf/cudf/core/udf/masked_lowering.py#L78) in a loop, but probably doable.

Contributor guide

Open the contributing guide

Assessment

This issue has not been assessed yet.

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.