Analyze replacement gofuzz with native Go fuzz testing

[antonioiubatti93]

/cc @RemiMattheyDoret following this suggestion

[antonioiubatti93]

@RemiMattheyDoret I gave it a try locally to understand how native fuzzing works. I must say I'm quite disappointed by how it works, but maybe I'm not employing it correctly. I think what we need here is just a unit test, not a fuzz test running continuously.
Reporting here what I did.

Fuzz test

Adapting the existing test:

func Fuzz_YearFraction_EqualDates(f *testing.F) {
	f.Add(0)

	origin := date.New(2018, time.January, 1)

	f.Fuzz(func(t *testing.T, days int) {
		from := origin.Add(days)

		for _, convention := range []Convention{
			ActualActual,
			ActualActualAFB,
			ActualThreeSixty,
			ActualThreeSixtyFiveFixed,
			ThirtyThreeSixtyEuropean,
			ThirtyThreeSixtyGerman,
			ThirtyThreeSixtyItalian,
			ThirtyThreeSixtyUS,
		} {
			convention := convention

			t.Run(convention.String(), func(t *testing.T) {
				t.Parallel()

				assert.Equal(t, 0.0, YearFraction(from, from, ActualActual))
			})
		}
	})
}

How to run

Fuzz test runs indefinitely (until failure, stopped or panic). To run the test (limited time with fuzztime, otherwise Ctrl+C works):

go test -race -fuzz=Fuzz_YearFraction_EqualDates -fuzztime=10s ./...

and an output should start appearing:

fuzz: elapsed: 0s, gathering baseline coverage: 0/7 completed
fuzz: elapsed: 0s, gathering baseline coverage: 7/7 completed, now fuzzing with 8 workers
fuzz: elapsed: 3s, execs: 6105 (2035/sec), new interesting: 4 (total: 11)
fuzz: elapsed: 6s, execs: 12173 (2023/sec), new interesting: 4 (total: 11)
fuzz: elapsed: 9s, execs: 18080 (1968/sec), new interesting: 4 (total: 11)
fuzz: elapsed: 10s, execs: 20012 (1736/sec), new interesting: 4 (total: 11)
PASS

[RemiMattheyDoret]

That's cool. The previous test created a single random value. This one indeed just fuzzes until the fuzztime is elapsed. I suspect all fuzz tests will have the same fuzz time (which is maybe not very practical), while with the previous version you could just build your own loop and fuzz for as long (or as many times) as you wish. I don't quite understand how seeding (seed corpus) work either. I am not able to judge whether it'd be better to transition to native Go fuzzing or not.