openjdk@8 1.8.0-432

[calvinit]

• Have you followed the guidelines for contributing?
• Have you ensured that your commits follow the commit style guide?
• Have you checked that there aren't other open pull requests for the same formula update/change?
• Have you built your formula locally with HOMEBREW_NO_INSTALL_FROM_API=1 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 HOMEBREW_NO_INSTALL_FROM_API=1 brew install --build-from-source <formula>)? If this is a new formula, does it pass brew audit --new <formula>?

[calvinit]

Why is the openjdk version used to test embulk here openjdk@21 instead of openjdk@8? Did the following check fail to work correctly?

on_arm do
  depends_on "openjdk@21"
end
on_intel do
  depends_on "openjdk@8"
end

I printed the content of the embulk script at bin/embulk on my Intel Mac, and it is as follows:

#!/bin/bash
export JAVA_HOME="${JAVA_HOME:-/usr/local/opt/openjdk@8/libexec/openjdk.jdk/Contents/Home}"
exec "${JAVA_HOME}/bin/java"  -jar "/usr/local/Cellar/embulk/0.11.5/libexec/embulk-0.11.5.jar" "$@"

It seems that the JAVA_HOME environment variable already has a value (which is /usr/local/opt/openjdk@21/libexec/openjdk.jdk/Contents/Home), so the subsequent openjdk@8 environment variable value is not being applied. Alternatively, the value of java_version might actually be 21 (java_version = Hardware::CPU.intel? ? "1.8" : "21"). Could this be a bug in the test-bot or brew? Because when testing embulk individually, there is no problem, but here, when it is tested as a dependency of openjdk@8 1.8.0-432, the problem arises.

P.S. def install of embulk.rb:

def install
  java_version = Hardware::CPU.intel? ? "1.8" : "21"
  libexec.install "embulk-#{version}.jar"
  bin.write_jar_script libexec/"embulk-#{version}.jar", "embulk", java_version:
end

[calvinit]

Help wanted! @cho-m
Why is the embulk test section unable to correctly retrieve the value of the JAVA_HOME environment variable?
#191470

[carlocab]

Ping @Homebrew/core -- @calvinit could use some help here. (I'll try to look myself later today.)

[calvinit]

Ping @Homebrew/core -- @calvinit could use some help here. (I'll try to look myself later today.)

Thank you, it would be the best.

I'll talk about the latest progress now: with the latest PR commit now, the build-from-source compilation on my macOS 15 Intel laptop is successful, but now the macOS 14 CI PR check is a bit problematic (in fact, it was okay before, but it didn't work after retrying because of the test embulk problem, and it feels like it may be related to upgrading Xcode).

The other one is the test embulk issue mentioned above, JAVA_HOME the wrong choice of openjdk@21 causes the problem (on the x86_64 architecture, it should choose openjdk@8, which is the Formula of this PR).

BTW: it looks like JNF is no longer a hindrance to compiling openjdk@8 on macOS 14+, is it possible to add CI check for macOS 15 x86_64 architecture as well as a dedicated bottle?

[Bo98]

Given the tbd file is still there in the SDK and it's just the headers that are missing - have you tried just -isystem the headers directory here rather than building the whole framework?

[calvinit]

I haven’t tried it yet. It sounds like a feasible approach, but I first need to find a source that can be downloaded to these headers.

[carlocab]

Suggested change

	DevelopmentTools.clang_build_version == 1600 &&
	ENV.append_to_cflags("-mllvm -enable-constraint-elimination=0")
	if DevelopmentTools.clang_build_version == 1600
	ENV.append_to_cflags("-mllvm -enable-constraint-elimination=0")
	end

Could we change this please?

[calvinit]

@carlocab Hi, I have tried both 2 all, but CI / tap_syntax (pull_request) check cannot pass.