openjdk 16

[jonchang]

This is still in release candidate stage but we might as well get to work fixing all the formulae. Development has moved to GitHub so we have the joy of maintaining 3 different repository systems for openjdk@8, openjdk@11 and openjdk.

While I'm here, start upstreaming some Linux changes.

• Have you followed the guidelines for contributing?
• Have you checked that there aren't other open pull requests for the same formula update/change?
• Have you built your formula locally with brew install --build-from-source <formula>, where <formula> is the name of the formula you're submitting?
• Is your test running fine brew test <formula>, where <formula> is the name of the formula you're submitting?
• Does your build pass brew audit --strict <formula> (after doing brew install <formula>)?

---

[jonchang]

Note to self: use openjdk/jdk pr 2200 for arm testing

[SMillerDev]

On ARM

/private/tmp/openjdk-20210306-55935-5nxkb5/jdk16-jdk-16-36/src/hotspot/share/runtime/globals.hpp:36:10: fatal error: 'globals_bsd_aarch64.hpp' file not found
436
#include OS_CPU_HEADER(globals)
437
         ^~~~~~~~~~~~~~~~~~~~~~

[Bo98]

JEP 391 did not make it into OpenJDK 16. If we still want to use the Apple Silicon changes, we'll need to incorporate it as a patch. openjdk/jdk#2200 contains the most recent changes in regards to this (though perhaps better to prepare a patch closer to release so we get the latest bugfixes that may be added to the PR between now and then). The branch on jdk-sandbox is a little old now.

[jonchang]

Yeah, this was more to check which of the revdeps would have issues. I'll prepare a ARM build after release. I'll probably have to pull one of our M1s out of rotation so I can build it interactively.

[jonchang]

brew livecheck openjdk --full-name --debug
brew install --build-from-source apache-forrest
brew install --build-from-source carrot2
brew install --build-from-source dafny
brew install --build-from-source emscripten
brew test --retry --verbose google-java-format
brew test --retry --verbose gradle
brew test --retry --verbose gradle-profiler
brew test --retry --verbose kotlin
brew install --build-from-source kotlin-language-server
brew install --build-from-source micronaut
brew test --retry --verbose micronaut
brew install --build-from-source pdftk-java
brew test --retry --verbose redpen
brew test --retry --verbose ringojs
brew install --build-from-source smali
brew install --build-from-source swagger-codegen
brew test --retry --verbose swagger2markup-cli

[Moisan]

emscripten has been bumped from 2.0.14 to 2.0.15 since the last CI run. It might fix itself upon the next rebase. 🤞

[Moisan]

micronaut harcodes the JDK versions they support, which does not include 16 at this time. Either we wait for their new release, or we fall back on openjdk@11.

[Moisan]

ringojs is failing because of its dependency on Rhino and this issue: mozilla/rhino#462

[mkfmnn]

JDK 16 went GA today: https://mail.openjdk.java.net/pipermail/announce/2021-March/000295.html

[cemo]

@jonchang would you please update this PR?

[Moisan]

build failures on 11.0:

• carrot2
• dafny
• dartsim
• kotlin-language-server
• micronaut
• openmodelica: openmodelica 1.17.0 #73671
• pdfpc
• pdftk-java
• smali
• swagger-codegen

test failures on 11.0:

• detekt
• google-java-format: google-java-format 1.10.0 #74524
• gradle: Seems like gradle 7.0 will support Java 16 Support Java 16 gradle/gradle#13481
• gradle-profiler
• kotlin
• ktlint: Trailing Whitespace causes format (-F) to fail the first time pinterest/ktlint#1116
• micronaut: Upgrade to Gradle 7  micronaut-projects/micronaut-core#5179
• redpen
• ringojs: depends on Rhino Illegal Reflective Operation mozilla/rhino#462
• swagger2markup-cli

[Moisan]

For ARM, openjdk doesn't build.

[SMillerDev]

Does anyone know if there is a patch to make OpenJDK 16 build on ARM? Or should we still be shipping the preview?

[Moisan]

There is currently a new openjdk16 release: jdk-16.0.1-ga

[Moisan]

Thanks @cho-m for the summary. I rebased the PR on top of master.

[cho-m]

ARM Failure:

• Commits for build not backported to JDK16. Will need to use JDK17-ea

Failures possibly related to JDK or Java dependencies:

• carrot2 - dependency doesn't support Gradle 7 - Gradle 7 compatibility palantir/gradle-consistent-versions#700
• druid - Test failure with zookeeper already running as process 70147
• wartremover - Test failure possibly from some version mismatch. May want to check if sbt needs to be rebuilt or if there is mismatch in Scala versions between compilation and runtime.
• kotlin kotlin 1.5.0 #76677 - Kotlin 1.4 issues with JDK16. EDIT: Kotlin updated to 1.5.
• gradle-profiler gradle-profiler: update for gradle 7.0 #76876 - Test failure at gradle-profiler --gradle-version 6.5

Attempting downgrade to JDK@11

• dafny - need triage... stacktrace appears to be Gradle-related
• ki - Upstream Kotlin 1.4.32 dependencies
• detekt - Upstream Kotlin 1.4 dependencies: Update Kotlin to 1.5 detekt/detekt#3718
• ktlint - Upstream Kotlin 1.4.31 dependencies
• redpen, swagger2markup-cli - dependency Pegdown (Deprecated) issue with JDK16
• rest-shell
• ringojs - triage from before: depends on Rhino Illegal Reflective Operation mozilla/rhino#462
• swagger-codegen - JDK16 failure. need to be fixed upstream [JAVA] java.lang.reflect.InaccessibleObjectException: Unable to make public boolean java.util.Collections$EmptyMap.isEmpty() accessible swagger-api/swagger-codegen#10966

JDK failures requiring formula updates:

• micronaut (Fix in this PR) - formula will need to be updated to 2.5.0 after JDK16 is added.
• google-java-format (Fix in this PR) - For JDK16, Homebrew Jar script will need --add-exports from https://github.com/google/google-java-format/releases/tag/v1.10.0

Separate failures that don't seem related to JDK:

• tree-sitter - PENDING ANALYSIS
• msc-generator, zile - Common Perl Binary mismatch for macOS 11.3
• bnfc - New Sphinx4 directory structure. man/bnfc.1 is probably man/1/bnfc.1
• pdfpc - Upstream issue Fix build with vala 0.52+ due to API change in gstreamer pdfpc/pdfpc#594 with newer vala. Vala 0.48.17 has Gst.Video.Info.from_caps but Vala 0.52+ removed it. Can consider patching build or just leave it.
• djl-serving djl-serving: fix sha256 mismatch #76668 - SHA mismatch
• mlpack mlpack: update stb headers #76835 - Resource missing: https://mlpack.org/files/stb-2.22/stb_image.h.
• openmodelica openmodelica: fix build with CMake 3.20+ #76836 - build issue with autoconf and cmake
• selenium-server-standalone selenium-server-standalone: refactor assert_true in test #76660 - Test failure using assert_true usage
• hbase - Resource hadoop-lzo github repo missing EDIT: GitHub repo restored

[carlocab]

Only gtkspell3 has an ARM bottle from the failures on ARM, as far as I can tell.

[cho-m]

@carlocab I think it is unrelated to OpenJDK. Failure is due to missing build dependency (maybe due to changes in another dependency). I opened PR fix in #78388

[cho-m]

I think we are good to merge at this point. Others on this thread can chime in for final review.

---

Big Sur:

Error: 2 failed steps!
brew install --build-from-source opensearch
brew install --build-from-source pdfpc

Catalina:

Error: 4 failed steps!
brew install --build-from-source opensearch
brew install --build-from-source pdfpc
brew test --retry --verbose subversion
brew test --retry --verbose subversion

Mojave

Error: 2 failed steps!
brew install --build-from-source opensearch
brew install --build-from-source subversion

[SMillerDev]

Looks good to me

[BrewTestBot]

@carlocab has triggered a merge.

[MikeMcQuaid]

Nice work here @jonchang and everyone 👏🏻!

[SMillerDev]

Issues on Linux: https://github.com/Homebrew/linuxbrew-core/runs/2710050156?check_suite_focus=true

[Toberumono]

@bdw429s This is a bit late, but my attempts at creating a PR for Apache Felix revealed a cascading number of reflection issues. Unfortunately, resorting to the --add-opens flags will be the only way to run on Java versions higher than 11 until Felix gets around to actually fixing their codebase.

In general, this means that anything depending on Apache Felix will need to remain pinned to Java 11 for the foreseeable future.

[bdw429s]

@Toberumono Thanks for giving it a go. I'm fine with pinning to Java 11 for now. It is the current LTS and hopefully the Felix devs will get with the program soon.