add tiger hash from https://gitlab.com/rawler/tiger

[myers]

Add the tiger hash.

The bulk of this is from Ulrik Mikaelsson but also work from Igor Raits and myself in upgrading to newer versions of the Digest crate.

@tarcieri suggested adding this crate to this repo so that it can be updated as needed.

[tarcieri]

Looks like you need to rustfmt

[tarcieri]

All of our other crates are MIT OR Apache-2.0. Is that acceptable?

[myers]

that's a good question. The original work is released under Apache 2.0. From reading https://law.stackexchange.com/questions/6081/can-i-bundle-mit-licensed-components-in-a-apache-2-0-licensed-project it seems like adding this under the MIT license is easy?

Should we ask Ulirk?

[tarcieri]

It would definitely be good to get the original author's blessing to republish it as MIT+Apache2.0

[myers]

I have emailed him. I started to make it look like it was under both licenses, but removed that.

[myers]

based on the failure on thumbv7em-none-eabi the byte-order crate doesn't work under nostd. Looks like what it does you can use byte-tools instead. I'll work on that.

[myers]

@tarcieri I think I have addressed everything you mentioned above.

If the license question is an issue, I could also go and resurrect #50

[tarcieri]

The license issue is pretty important as we try to ensure our crates all have consistent licensing and a clear IPR story.

We can wait to hear from the author or you can revive #50.

Perhaps you can benchmark the two and see if there's a compelling reason other than licensing to go with one over the other?

[myers]

From: Ulrik Mikaelsson
Date: Fri, 5 Feb 2021 07:35:53 +0100
Subject: Re: Relicense rust tiger under MIT?

Fun to hear you're using it! I'm ok with licensing it under MIT, but i also
got notice the the PR is now closed in favor of other PR, so not sure it
matters?

I will make the changes to have it listed as under MIT OR Apache 2.0

[myers]

benchmark of this PR's tiger crate:

$ cd tiger
$ cargo bench
[...]
test bench1_10 ... bench: 20 ns/iter (+/- 3) = 500 MB/s
test bench2_100 ... bench: 158 ns/iter (+/- 17) = 632 MB/s
test bench3_1000 ... bench: 1,678 ns/iter (+/- 240) = 595 MB/s
test bench4_10000 ... bench: 15,735 ns/iter (+/- 2,818) = 635 MB/s
test result: ok. 0 passed; 0 failed; 0 ignored; 4 measured; 0 filtered out; finished in 11.33s

[myers]

I have spent 20 minutes to try to get #50 to compile and pass tests so that I can benchmark, but it's failing tests.

I propose this PR is ready for merging.

[tarcieri]

Looks good, thank you!

I'll cut a 0.1 after merging.

[tarcieri]

It's released: https://crates.io/crates/tiger