Compilações Incrementais e Entradas e Saídas de Tarefas

The fastest task is the one Gradle does not run. Incremental builds let Gradle skip tasks whose inputs and outputs have not changed since the last run.

When a task reports UP-TO-DATE, Gradle compared the current inputs/outputs against a stored snapshot and found no change, so it skipped execution.

For Gradle to track a custom task, annotate its inputs. Files, directories, and simple properties all qualify.

Outputs tell Gradle what the task produces. If an output is deleted or changed, the task reruns.

Non-file inputs use @Input. Changing the value invalidates the cached result.

A task with no declared inputs and outputs has nothing to compare, so Gradle runs it every time. Always declare them to gain incrementality.

Advanced tasks process only the changed files using InputChanges, instead of reprocessing everything.

Correctly declared inputs/outputs also unlock the build cache: identical inputs across machines reuse stored outputs.

Use the --info flag to see why a task was not UP-TO-DATE. Gradle prints which input or output changed.

Watch out for inputs that change every run:

• Timestamps embedded in outputs
• Absolute paths in inputs
• Reading system time or random values

These break incrementality.

Use path sensitivity to ignore irrelevant differences, e.g. classpath order, so equivalent inputs hash the same.

You learned incremental builds:

• UP-TO-DATE means inputs/outputs were unchanged
• Annotate with @Input, @InputFile, @OutputFile
• Undeclared tasks always run
• Avoid timestamps and absolute paths that break tracking
• Correct declarations also enable the build cache