I don't think that (grep) is really possible. I still want to see log output when a test fails. That's the great thing about t.Log. I don't think there's any way to grep such that I still get log output. Note that I am mainly thinking of using this on a CI server. Locally, I may be able to come up with a grep call that would strip all logging... But I'd have to rerun tests to get log output once I notice some failing. And that's not really an option on a CI server.

I'm not completely opposed but it does sound a lot like featuritis to me. Grep is your friend. Let other tools do some of the lifting.

How about this?

zapf(~/src/tt) % go test -v
=== RUN   TestSucceeding
--- PASS: TestSucceeding (0.00s)
=== RUN   TestFailing
--- FAIL: TestFailing (0.00s)
	test_test.go:10: whoops, about to fail
	test_test.go:11: ker-bham!
FAIL
exit status 1
FAIL	tt	0.006s
zapf(~/src/tt) % go test -v 2>&1 | grep "^=== RUN"
=== RUN   TestSucceeding
=== RUN   TestFailing

Or this

% go test -list Test
TestSucceeding
TestFailing
ok  	tt	0.005s

The former requires me to run my tests twice and still forces me to wade through masses of logs to find log output for the failing tests. The second mostly works (if I then run go test without -v), it at least gives me the list of tests that'll run, but doesn't give me execution times of individual tests.

I just wish the two features were separated - the "always show logging" and "show individual tests being run".

Once we do machine-readable test output you can split this as many different ways as you want. Let's plan for that instead of adding more control knobs here.