Improve tests

[mgeier]

• we shouldn't write a file with PySoundFile and then test reading it with PySoundFile. If there is a systematic error in reading and writing, we won't find it
• we shouldn't write a file in each test case. It would be better to create one (or few) very small test file(s) that we ship with the repo (just a few hundred bytes). We should only write files when we're testing writing files.
• we shouldn't fill the test file with copies of the same number. The samples should be different.

[mgeier]

Apart from this, there also some formal problems:

$ pep8 test_pysoundfile.py 
test_pysoundfile.py:7:1: E302 expected 2 blank lines, found 1
test_pysoundfile.py:14:80: E501 line too long (84 > 79 characters)
test_pysoundfile.py:39:80: E501 line too long (83 > 79 characters)
test_pysoundfile.py:45:80: E501 line too long (85 > 79 characters)
test_pysoundfile.py:46:80: E501 line too long (84 > 79 characters)
test_pysoundfile.py:129:80: E501 line too long (81 > 79 characters)
test_pysoundfile.py:145:33: E231 missing whitespace after ','
test_pysoundfile.py:160:80: E501 line too long (84 > 79 characters)
test_pysoundfile.py:229:80: E501 line too long (80 > 79 characters)
test_pysoundfile.py:235:80: E501 line too long (84 > 79 characters)
test_pysoundfile.py:265:80: E501 line too long (85 > 79 characters)
test_pysoundfile.py:271:49: E231 missing whitespace after ','
test_pysoundfile.py:272:46: E231 missing whitespace after ','
test_pysoundfile.py:275:80: E501 line too long (94 > 79 characters)
test_pysoundfile.py:290:80: E501 line too long (84 > 79 characters)
test_pysoundfile.py:297:1: E302 expected 2 blank lines, found 1
test_pysoundfile.py:301:33: E231 missing whitespace after ','
test_pysoundfile.py:311:37: E231 missing whitespace after ','
test_pysoundfile.py:316:32: E231 missing whitespace after ','
test_pysoundfile.py:324:33: E231 missing whitespace after ','
test_pysoundfile.py:324:47: E261 at least two spaces before inline comment
test_pysoundfile.py:335:33: E231 missing whitespace after ','

[bastibe]

These are very valid points. Thank you!

[mgeier]

I think the tests about precision are not really necessary because they only test libsndfile.
I think it would be better to check for exact equality without the PCM <--> float conversion.

So either make a float test file and check with float32/float64 or make a PCM test file and test with int16/int32.
Or both.

It would be probably best to make a floating point test file with values which are accurately representable in both decimal and binary.
Something like [1.0, 0.5, 0.0, -0.5, -1.0], probably?
In stereo this could be [[1.0, -1.0], [0.5, -0.5], [0.0, 0.0], [-0.5, 0.5], [-1.0, 1.0]].
I guess 5 frames would be enough for most test cases.

[bastibe]

I am closing this issue. Discussion continues in #48 .