This hash should be 1xpblx405cp3mv0vrhcqwjlxvrhgmc77mxbvcy93srxja3qai1af

Oh. I had done nix-prefetch-url --unpack https://github.com/unisonweb/unison/releases/download/release/M1j/unison-linux64.tar.gz, which resulted in a different hash. When should you use --unpack and when shouldn't you? I tried finding more information on this and couldn't find much. I think that I did the same for the Darwin file, so maybe that needs to change too? But it worked fine for me when I ran it locally.

Oh never mind, looking at my shell history I didn't do the same for Darwin, so it's probably right. Still if someone could shed some light on when to --unpack and when not to, I'd appreciate it.

@ceedubs --unpack is a hash of the content. You will need to use --unpack to get hashes for fetchzip, since that hashes the content. Not using --unpack is a hash of the archive.

@virusdave @worldofpeace
Sorry to butt in, but could/would someone help me understand the nix hashing here better? When I try to compute it using shasum (which is what we use for the ucm homebrew formula), I get a different result from any of the above:

$ wget https://github.com/unisonweb/unison/releases/download/release/M1j/unison-linux64.tar.gz
$ shasum -a 256 unison-linux64.tar.gz 
4e85a8f050b2673d92677bf57a0eab0fe6dda9e498c1bcc1aee3b20248a7ebf6  unison-linux64.tar.gz
$ shasum -a 1 unison-linux64.tar.gz 
17d7b14f0eed55eb5edc31088db67ebc77168935  unison-linux64.tar.gz

Any ideas why? Thanks in advance.

Oh, hm, I guess it has to do with the encoding of the hash; shasum is using base16, and nix is using... base32, maybe?

No, it doesn't seem to match according to any base32 alphabet I can find (rfc4648 base32, z-base-32, crockford's, base32hex, geohash). 😕

I guess nix uses its own alphabet, though I still can't seem to make the hash bits match up; but maybe it's enough to say that the appropriate hash comes from the nix tools (and not from elsewhere) and that's okay :)

@virusdave @worldofpeace
Sorry to butt in, but could/would someone help me understand the nix hashing here better? When I try to compute it using shasum (which is what we use for the ucm homebrew formula), I get a different result from any of the above:

$ wget https://github.com/unisonweb/unison/releases/download/release/M1j/unison-linux64.tar.gz
$ shasum -a 256 unison-linux64.tar.gz 
4e85a8f050b2673d92677bf57a0eab0fe6dda9e498c1bcc1aee3b20248a7ebf6  unison-linux64.tar.gz

$ nix-hash --to-base16 --type sha256 1xpblx405cp3mv0vrhcqwjlxvrhgmc77mxbvcy93srxja3qai1af

4e85a8f050b2673d92677bf57a0eab0fe6dda9e498c1bcc1aee3b20248a7ebf6


$ nix-hash --to-base32 --type sha256 4e85a8f050b2673d92677bf57a0eab0fe6dda9e498c1bcc1aee3b20248a7ebf6 

1xpblx405cp3mv0vrhcqwjlxvrhgmc77mxbvcy93srxja3qai1af

They match :)