test_cratesio_hash fails on systems that are not little-endian 64-bit

[infinity0]

Problem

Not directly cargo's problem, but test_cratesio_hash is how I discovered the problem: rust-lang/rust#90230

Steps

No response

Possible Solution(s)

No response

Notes

No response

Version

No response

[infinity0]

It may be reasonable for rustc to not fix the underlying code of SipHasher as it is already a deprecated API; however from cargo's point of view it should decide what "StableHasher" actually means as the current implementation is bit-width- and endian-dependent and goes against intuitive notions of "Stable". In any case test_cratesio_hash should be adapted/fixed to address this situation.

For the time being I will just ignore the test failure in Debian, as the behaviour has apparently existed since 1.44 (cargo 0.45) without major complaints that I'm aware of.

[infinity0]

CC @alexcrichton who wrote the test

[infinity0]

rust-lang/rust#90230 was closed in favour of rust-lang/rust#74215

[alexcrichton]

If this is little-endian dependent then seems like it's fine to rewrite the test or update it not be so.

[hkratz]

As far as I understand the test comment it is not important that those hashes are the same across platforms so I have adjusted the expected values in #10011.

[infinity0]

The naming "StableHasher" is misleading though.