probably a stupid question but why do we expect 3 here (instead of 4)?

Because we're getting contents in range get_sha1s[1] to get_sha1s[-1] inclusive, so that's all the contents (4) minus the first one.

oh yes indeed.

Wouldn't it be useful (?) to raise the max_size and check the content_get_range against all situations wrt range limits: the beginning, middle and end of the range (like range(0, 2), range(n, n+2) and range(-3, -1) )?

That's what the three tests are for (test_generate_content_get_range_start, test_generate_content_get_range_end, test_origin_get_range). I did not want to do a single test for all of them, because it means writing the same logic in the test as in the real code, so testing makes less sense

Oh you mean I must read the whole diff when I review it? what a bama.

So now, why does the range_start test uses sha1[1] as start value?

range_start tests that nothing before the start is returned, range_end tests that nothing after the end is returned.

Eh, you're right, I could merge these two tests