b46c76ea89
Buck runs tests in parallel based on the targets declared in BUCK files. Making each test its own target allows the tests to run in parallel in separate JVMs. Parallel execution enables slow acceptance tests to take less wall clock time when multiple cores are available. Acceptance tests leak an entire server environment for every @Test method run. Fragmenting the tests into units allows them to be garbage collected by the operating system when the test JVM dies. This avoids stressing the Java GC with lots of garbage as more tests pile into the finite heap space. With this change and the newer version of buck, a clean build and execution of all tests takes my laptop only 1m25s to execute when running on battery. Change-Id: Ia3f689a9e6ca6c620026cc720ccca924387e1363
10 lines
191 B
Python
10 lines
191 B
Python
java_library(
|
|
name = 'launcher',
|
|
srcs = glob(['src/main/java/**/*.java']),
|
|
visibility = [
|
|
'//gerrit-acceptance-tests/...',
|
|
'//gerrit-main:main_lib',
|
|
'//gerrit-pgm:',
|
|
],
|
|
)
|