Split slower tests to a separate test target
If a test binary didn't change in a CL, ATP will skip running it. Right
now, all update engine unittests are packed into a single target called
update_engine_unittests. This is bad, as any changes in code will cause
this test binary to change, triggering re-run of all test cases.
This CL splits some of the HTTP related testcases to a separate test
target. We rarely change HTTP related code, and these test cases take a
long time to run. For majority of code changes that don't touch HTTP,
ATP should be able to skip running HTTP testcases and save us time.
Test runtime:
1. http unitests: 3min
2. all other update_engine unitests: 3min
Test: th
Change-Id: Ieb5bf986a5fedf38710d96a4a9c974e5a1c1491d
diff --git a/TEST_MAPPING b/TEST_MAPPING
index 93777d4..3a9a238 100644
--- a/TEST_MAPPING
+++ b/TEST_MAPPING
@@ -2,6 +2,9 @@
"presubmit": [
{
"name": "update_engine_unittests"
+ },
+ {
+ "name": "update_engine_http_unittests"
}
]
}