Skip to content

docs: add contributor guide for bringing up a new Spark version#4133

Draft
andygrove wants to merge 3 commits intoapache:mainfrom
andygrove:add-spark-version-guide
Draft

docs: add contributor guide for bringing up a new Spark version#4133
andygrove wants to merge 3 commits intoapache:mainfrom
andygrove:add-spark-version-guide

Conversation

@andygrove
Copy link
Copy Markdown
Member

Which issue does this PR close?

N/A. Documentation-only addition; no tracking issue.

Rationale for this change

Bringing up a new Apache Spark version in Comet (recent examples: 4.0, 4.1, 4.2-preview) touches the build, the shim layer, CI, and three different test surfaces (Comet's JVM tests, Spark's SQL tests, and TPC-DS plan stability). When that work is bundled into one PR the diff is hard to review and harder to bisect, and contributors who have not done a Spark bring-up before have to reverse-engineer the recipe from prior PRs.

This page captures the recipe explicitly so the next bring-up can follow a known path.

What changes are included in this PR?

  • New page docs/source/contributor-guide/adding_a_new_spark_version.md describing a staged-PR workflow:
    1. PR 1: Maven profile, shims (including the spark-4.x shared / spark-X.Y per-version layering), version-detection helper, compile-only CI job.
    2. PR 2: Enable Comet's JVM tests under the new profile; skip failing tests with assume() calls linked to GitHub issues.
    3. PR 3: Enable Spark SQL tests and plan stability under the new profile; same skip-with-linked-issue rule.
    4. Follow-up PRs: one issue (or one cluster) per PR, removing the corresponding skip as part of the fix.
  • Cross-reference added to docs/source/contributor-guide/spark-sql-tests.md so contributors who land on the existing diff-creation section get pointed at the broader staged plan first.
  • New page added to the contributor guide toctree in index.md.
  • The diff-creation mechanics already covered in spark-sql-tests.md are intentionally not duplicated; the new page links to that section instead.

How are these changes tested?

Documentation only; no code changes. Verified the toctree entry renders by inspecting index.md and confirmed the cross-references resolve to existing files in the contributor guide.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant