Replay Gain Test: Leaving repo unclean

[georgev93]

It looks like the last test (testing the --replaygain feature) does not leave the repo clean:

ffmpeg-normalize/test/test.py

Lines 479 to 489 in b25a373

	def test_replaygain(self):
	for file in [
	"test/test.mp4",
	"test/test.mp3",
	"test/test.ogg",
	"test/test.opus",
	]:
	original_mtime = os.path.getmtime(file)
	ffmpeg_normalize_call([file, "--replaygain"])
	assert os.path.isfile(file)
	assert os.path.getmtime(file) > original_mtime

After running this test, git flags a change of the test/test.mp4 file, since the replay gain tag of that mp4 changes from 7.44dB to 7.93dB.

Not sure what that indicates (as I'm not fully spun up on what a "replay gain" means or what this feature is supposed to do), but the repo shouldn't get dirty simply by running tests. Perhaps commit the mp4 post-test?

[slhck]

Good catch! The ReplayGain feature modifies the file metadata, so after the test, the file should just be reverted to keep the repo clean.