fix(rapids-artifact-name): handling for "true" pure artifacts vs. cuda-dependent pure artifacts#254
Merged
Conversation
…a-dependent pure artifacts
gforsyth
added
improvement
KyleFromNVIDIA
approved these changes
Apr 29, 2026
Contributor
Author
|
/merge |
gforsyth
added a commit
to gforsyth/gha-tools
that referenced
this pull request
Apr 29, 2026
… vs. cuda-dependent pure artifacts (rapidsai#254)" This reverts commit 2a51afa.
rapids-bot Bot
pushed a commit
that referenced
this pull request
May 6, 2026
…e `$(arch)` (#255) - **refactor(rapids-artifact-name): add explicit `--arch` flag to override `$(arch)`** - **Revert "fix(rapids-artifact-name): handling for "true" pure artifacts vs. cuda-dependent pure artifacts (#254)"** Sorry for the churn on this. This PR reverts the change in #254 for a few reasons: 1. I was wrong about the categories of "purity" for artifacts (or not right enough) -- `distributed_ucxx` has a dependency on cuda version (or is built for both cuda 12 and cuda 13, at least) but is CPU arch independent. 2. My solution for the categories of "purity" made the inclusion of the CPU architecture in the artifact name a side-effect, which is bad. So, instead, after reverting that change, I'm adding an additional (and optional) `--arch <value>` flag, that, if used, overrides the value of `$(arch)` with the user-provided string. This could be useful for downloading artifacts with a different architecture than the system a developer is currently using, and can also be used to set an arbitrary value for `arch` for the cases like `distributed_ucxx` where we build on `x86_64` but use the artifacts built there on both `x64_64` and `aarch64` systems. (This is true for anythign that doesn't have compiled code but does have CUDA-version-specific dependencies, like `nx-cugraph` and `cuopt-server`) Authors: - Gil Forsyth (https://github.com/gforsyth) Approvers: - James Lamb (https://github.com/jameslamb) URL: #255
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
xref rapidai/build-planning#270
Working on rapidsai/ucxx#642 which raised the issue of our different types of "pure" artifacts -- those that are truly Python-only and so have no cuda or architecture dependencies, and those that are agnostic to Python version, but still have cuda dependencies.
Adding handling here, so that passing
--purewithout--cudawill elide the{arch}part of the artifact name. If we use--pureand--cuda, then{arch}will be included.If
--pureisn't used,{arch}is always included.