[ptweezy]

When IGzipFile.__init__() doesn't set self.compress, GzipFile compresses with zlib instead, and no test checked for that. The new test replaces isal_zlib.compressobj() with a wrapper that records its arguments. It checks that opening an IGzipFile for writing calls it once with the given level and raw deflate settings, and that f.compress is an isal_zlib.Compress.

To test it, I removed that line, and in another copy, always passed level 2 to compressobj(). The old tests in tests/test_igzip.py and the gzip compliance tests pass in both cases. The new test fails at all four levels with the first change, and at levels 0, 1, and 3 with the second.

This PR changes only tests, so it has no CHANGELOG entry.

Checklist

• Pull request details were added to CHANGELOG.rst
• Documentation was updated (if needed)

[rhpvorderman]

Why not assert isinstance(f.compress, type(isal_zlib.decompressobj()) ?

EDIT: Oh I see, this also tests whether the correct level is passed.

Rather than byte for byte testing I prefer to have something that tests the actual parameters rather than the result.

[ptweezy]

Changed the test to check the parameters. It now replaces isal_zlib.compressobj with a wrapper that records its arguments. It checks that IGzipFile calls it once with the given level and raw deflate settings, and that f.compress is an isal_zlib.Compress.

[rhpvorderman]

LGTM!