Optimize build: Remove shadowJar from build task

[cbb330]

Summary

Remove shadowJar from build task to speed up development builds by 56%.

The build.dependsOn shadowJar was explicitly added but is unnecessary because the Shadow plugin's maven-publish integration already triggers shadowJar when running publish. CI workflows are unaffected since ./gradlew publish runs before Docker builds.

Build time improvement (./gradlew clean build -x test):

	Time
Before	314s (5m 13s)
After	137s (2m 16s)
Improvement	-177s (56% faster)

Changes

• Client-facing API Changes
• Internal API Changes
• Bug Fixes
• New Features
• Performance Improvements
• Code Style
• Refactoring
• Documentation
• Tests

Performance Improvements

Removed tasks.build.dependsOn tasks.shadowJar from:

• buildSrc/src/main/groovy/openhouse.apps-spark-common.gradle
• tables-test-fixtures/tables-test-fixtures-iceberg-1.2/build.gradle
• tables-test-fixtures/tables-test-fixtures-iceberg-1.5/build.gradle

Why this is safe:

• publish task already triggers shadowJar via Shadow plugin's maven-publish integration
• CI workflow runs ./gradlew publish before Docker builds
• Tests that depend on configuration: 'shadow' still trigger shadowJar for their dependencies

Before (shadowJar runs on every build):

> Task :apps:openhouse-spark-apps_2.12:shadowJar
> Task :apps:openhouse-spark-apps-1.5_2.12:shadowJar
> Task :tables-test-fixtures:tables-test-fixtures_2.12:shadowJar
...
BUILD SUCCESSFUL in 5m 13s

After (shadowJar only runs on publish):

BUILD SUCCESSFUL in 2m 16s
250 actionable tasks: 244 executed, 6 up-to-date

Testing Done

• Manually Tested on local docker setup. Please include commands ran, and their output.
• Added new tests for the changes made.
• Updated existing tests to reflect the changes made.
• No tests added or updated. Please explain why. If unsure, please feel free to ask for help.
• Some other form of testing like staging or soak time in production. Please explain.

Manual Testing

1. Verified ./gradlew clean build -x test no longer runs shadowJar tasks
2. Verified ./gradlew publish --dry-run still triggers shadowJar
3. Measured before/after build times

No Tests Added

This is a build infrastructure change that doesn't affect runtime behavior.

Additional Information

• Breaking Changes
• Deprecations
• Large PR broken into smaller PRs, and PR plan linked in the description.

🤖 Generated with Claude Code

[jiang95-dev]

The apps uber jar is needed for the local docker. We need to update the README to add gradle publish before building dockers. I think it's ok to remove shadow jar for tables-test-fixures, but keep the apps to minimze the build steps for users.

[cbb330]

@jiang95-dev You are correct; this changes the workflow from ./gradlew build to ./gradlew build publish. However, this aligns with the default behavior of the shadowJar plugin in our other repos.

Mitigation Plan:

I will add a "heads up" notice and update the documentation in this PR.

I can also modify the Docker build command to throw a specific exception if the JAR is missing, explicitly instructing the user to run ./gradlew publish.

Motivation: Decoupling shadowJar from the app build significantly reduces build times. Since the standard build runs frequently (CI, post-merge, and ELR processes), the cumulative time savings are substantial.

[jiang95-dev]

Agree with the motivation. Can you update the readme?