Skip to content

fix(openjdk): restore temporary jdk 28 ea alias#471

Merged
jdx merged 1 commit into
mainfrom
codex/temp-openjdk-28-clean-alias
Jun 10, 2026
Merged

fix(openjdk): restore temporary jdk 28 ea alias#471
jdx merged 1 commit into
mainfrom
codex/temp-openjdk-28-clean-alias

Conversation

@jdx

@jdx jdx commented Jun 10, 2026

Copy link
Copy Markdown
Owner

Summary

  • add clean temporary OpenJDK 28 EA alias rows for Linux x64 and macOS arm64
  • use version = 28.0.0-ea so mise can resolve the exact e2e request
  • keep the reverted revert: back out sqlite data updates #469 baseline free of features: [""] corruption

Why

core/test_java_slow asks for java[release_type=ea]@openjdk-28.0.0-ea. After #469 reverted the bad generated data, that alias disappeared again. This restores the minimal clean rows needed for release CI and local macOS arm64 validation while a durable crawler/resolver fix is handled separately.

Validation

  • jq empty public/api/jvm/ea/linux/x86_64.json public/api/jvm/ea/macosx/aarch64.json
  • git diff --check

This PR was generated by an AI coding assistant.


Note

Low Risk
Static JVM catalog JSON only; no runtime or auth changes, aside from which EA JDK URL mise can install.

Overview
Restores OpenJDK 28 EA catalog entries that were lost when bad generated JVM data was reverted in #469.

Adds one JDK row each to public/api/jvm/ea/linux/x86_64.json and public/api/jvm/ea/macosx/aarch64.json, with vendor: openjdk, version / java_version: 28.0.0-ea, checksums, and download URLs from download.java.net. Rows use empty features arrays (no corrupted features: [""]). Both files also get a proper closing ] newline.

This lets mise resolve java[release_type=ea]@openjdk-28.0.0-ea for release CI (core/test_java_slow) and local macOS arm64 checks until crawler/resolver work lands.

Reviewed by Cursor Bugbot for commit 541722e. Bugbot is set up for automated code reviews on this repo. Configure here.

Summary by CodeRabbit

  • Chores
    • Added OpenJDK 28.0.0-ea Early Access builds for Linux x86_64 and macOS ARM64 with checksums and download URLs.

@coderabbitai

coderabbitai Bot commented Jun 10, 2026

Copy link
Copy Markdown

Review Change Stack

📝 Walkthrough

Walkthrough

This PR adds OpenJDK 28.0.0-ea entries to two platform-specific early-access JDK manifests: Linux x64 and macOS ARM64. Each entry includes checksum, creation timestamp, metadata fields, and download URL.

Changes

JDK 28 EA Platform Manifests

Layer / File(s) Summary
Add OpenJDK 28.0.0-ea to platform-specific manifests
public/api/jvm/ea/linux/x86_64.json, public/api/jvm/ea/macosx/aarch64.json
OpenJDK 28.0.0-ea entries with sha256 checksums, created_at timestamps, and download URLs are added to Linux x64 and macOS ARM64 early-access manifests.

Estimated code review effort

🎯 1 (Trivial) | ⏱️ ~3 minutes

Possibly related PRs

  • jdx/mise-java#463: Updates the OpenJDK crawler to include the JDK 28 page so that 28 EA artifacts can be discovered and published to these manifests.
  • jdx/mise-java#469: A revert that removes OpenJDK 28 EA entries from public/api/jvm/ea/linux/x86_64.json, directly overlapping with this PR's additions.

Poem

🐰 Oh, what a joy to add EA builds anew,
Version twenty-eight for Linux and Mac too,
Checksums dance in JSON's neat array,
Early access kernels brighten the day!

🚥 Pre-merge checks | ✅ 5
✅ Passed checks (5 passed)
Check name Status Explanation
Title check ✅ Passed The title directly describes the main change: restoring temporary OpenJDK 28 EA alias entries in the JSON manifests, which matches the file modifications shown in the raw summary.
Docstring Coverage ✅ Passed No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check.
Linked Issues check ✅ Passed Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check ✅ Passed Check skipped because no linked issues were found for this pull request.
Description Check ✅ Passed Check skipped - CodeRabbit’s high-level summary is enabled.

✏️ Tip: You can configure your own custom pre-merge checks in the settings.


Comment @coderabbitai help to get the list of available commands and usage tips.

@jdx jdx merged commit 4065b9e into main Jun 10, 2026
4 of 5 checks passed
@jdx jdx deleted the codex/temp-openjdk-28-clean-alias branch June 10, 2026 21:15
@greptile-apps

greptile-apps Bot commented Jun 10, 2026

Copy link
Copy Markdown

Greptile Summary

This PR manually appends two OpenJDK 28 EA build-1 entries (Linux x64 and macOS aarch64) to the EA JVM index files so the core/test_java_slow CI job can resolve java[release_type=ea]@openjdk-28.0.0-ea after the previous generated data was reverted in #469.

  • Both entries use features: [] (correct empty array), match the existing JSON schema, and fix the pre-existing missing-newline at the end of each file.
  • The new records hard-code build ea+1; this is explicitly acknowledged as a temporary measure until a crawler/resolver fix lands separately.
  • The SHA-256 checksums are manually supplied and the PR's stated validation steps (jq empty, git diff --check) do not confirm their correctness — they should be cross-checked against the Oracle-published .sha256 sidecar files.

Confidence Score: 3/5

The JSON structure and schema are correct, but the SHA-256 checksums are manually crafted and not verifiably confirmed in the PR — a wrong hash will break every installation that touches these entries.

Both files are well-formed and schema-consistent. The only unresolved concern is whether the two hand-written SHA-256 hashes match the actual Oracle-published values; if either is wrong, every mise install using these entries will fail with a checksum error, which is a silent breakage for end users.

Both public/api/jvm/ea/linux/x86_64.json and public/api/jvm/ea/macosx/aarch64.json — specifically the checksum fields of the newly added entries.

Important Files Changed

Filename Overview
public/api/jvm/ea/linux/x86_64.json Appends a single OpenJDK 28 EA build-1 entry for Linux x64 with sha256 checksum; also fixes the missing trailing newline.
public/api/jvm/ea/macosx/aarch64.json Appends a single OpenJDK 28 EA build-1 entry for macOS aarch64 with sha256 checksum; also fixes the missing trailing newline.

Fix All in Claude Code

Reviews (1): Last reviewed commit: "fix(openjdk): restore temporary jdk 28 e..." | Re-trigger Greptile

"image_type": "jdk",
"java_version": "28.0.0-ea",
"jvm_impl": "hotspot",
"url": "https://download.java.net/java/early_access/jdk28/1/GPL/openjdk-28-ea+1_linux-x64_bin.tar.gz",

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

P1 Manually-crafted checksums are unverified in the PR

The checksums sha256:d9b2b25f… (Linux) and sha256:bd5590d5… (macOS) were hand-written and the stated validation (jq empty, git diff --check) only confirms valid JSON and whitespace — it does not confirm the hash values are correct. A wrong checksum will cause every mise install attempt to fail with a checksum-mismatch error. The authoritative SHA-256 values are published by Oracle at https://download.java.net/java/early_access/jdk28/1/GPL/openjdk-28-ea+1_linux-x64_bin.tar.gz.sha256 (and the equivalent macOS URL) — please confirm these were fetched from there.

Fix in Claude Code

"vendor": "sapmachine",
"version": "13.0.2-ea.1"
},
{

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

P2 created_at lacks sub-second precision present in all other entries

Every other record in these files uses a microsecond-precision timestamp (e.g. "2025-03-28T22:04:28.081056"), while the new entries use only second precision ("2026-06-10T20:37:19"). If any downstream code sorts or deduplicates on created_at in combination with other fields, the inconsistent format could cause unexpected ordering. This applies to the macOS entry as well.

Fix in Claude Code

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