Skip to content

[FEA] use binary search to advance to the next cudf commit in submodule sync #2593

Open
@gerashegalov

Description

@gerashegalov

Is your feature request related to a problem? Please describe.

submodule-sync.sh is all-or-nothing currently. Recently there we ran into two commits breaking spark-rapids-jni

rapidsai/cudf@918266a
rapidsai/cudf@6acd33d

In general there could be a lot of commits between breaking commit A and a more recent breaking commit B. If A is fixed and B is not, submodule-sync would fail to advance the submodule missing on all the commits including B^

Describe the solution you'd like
It would be great to use binary search via git bisect to find the most recent advance-able cudf commit

Added benefit: output the earliest cudf commit breaking spark-rapids-jni

Describe alternatives you've considered
Keep current behavior of waiting until the most recent commit is workable

Additional context

Metadata

Metadata

Assignees

No one assigned

    Labels

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions