我认为这不是大多数人想要的。如果你需要，你可以自己在你的t.Log通话中添加时间戳。

就其价值而言，我认为这也是一个有用的功能，原因与@vimalk78提供的原因相同。如果我们有一个t.SetPrefix函数(类似于log包实现的方式),那将很酷。

同意，一直使用时间戳，并宁愿不要单独的日志包/在每个日志语句中包含一个TS片段。

I personally think this would be nice by default; most of the output from go test already includes relative timestamps, like how long a test ran for, or how long the entire package took to test. Here's an example of how relative timestamps on logs could look like:

--- FAIL: TestGetValue (1.50s)
    0.30s - main_test.go:9: this is log statement
    0.60s - main_test.go:11: this is error log statement
    0.90s - main_test.go:12: expected 100
FAIL
exit status 1
FAIL    github.com/vimalk78/test-testing        1.600s

cmd/go's TestScript does something similar; each section of a test script has a comment, and the output includes how long that section took to run (cc @bcmills):

# Verify that the go.mod files for both modules in the workspace are tidy,
# and add missing go.sum entries as needed. (0.032s)
> cp go.mod go.mod.orig
> go mod tidy
[...]
# Now remove the module dependencies from the module cache.
# Because one module upgrades a transitive dependency needed by another,
# listing the modules in the workspace should error out. (0.005s)
> go clean -modcache
> env GOPROXY=off

When I write long tests in Go, such as integration tests, I often want similar timestamps or durations to be able to tell what's taking most of the time. It's also useful with go test -v -run OneTest , so that the logs being streamed out can give a better idea of how the time is being spent.
Like @ianlancetaylor says, I can always add timestamps or elapsed durations to log calls myself, though at this point I've done that in a handful of different places, and I don't see a reason why we shouldn't do it by default :) I reckon I would generally find this useful for any test which takes more than a tens of milliseconds and prints a few logs.
Worth noting that, with go test -json , log lines get printed as "output" actions with a timestamp, but that's not very friendly for humans.

为了记录，我目前的解决方法是：

func TestFoo(t *testing.T) {
	start := time.Now()
	logf := func(format string, args ...any) {
		t.Logf("%.2fs: "+format, append([]any{time.Since(start).Seconds()}, args...)...)
	}
	[...]
	logf("cloning %s", ghSourceRepo)

带有 go test -v 的日志显示如下：

=== RUN   TestFoo
    foo_test.go:34: 0.00s: cloning mvdan/test-repo
    foo_test.go:34: 0.86s: committing a new file
    foo_test.go:34: 0.86s: pushing to mvdan/test-repo

就其价值而言，我在CL 405714中的cmd/go脚本测试中添加了绝对时间戳。然而，我只在运行开始时添加了一个时间戳：开始时的时间戳与测试的运行时间相结合，给出了一个有界的窗口，其中测试的事件发生，这是我真正关心的目的(诊断由非一致外部服务中的错误引起的测试失败)。