feat(stdlibs): import crypto/sha256

[albttx]

All credit to @r3v4s and his PR #570

Description

Import function Sum256 from package crypto/sha256 .

My goal is to implement a merkle-airdrop contract, so i need Sum256 function.

How has this been tested?

see unit test TestSha256Sum

[r3v4s]

Hello @albttx, thx for you comment back on #570.

I have question for this pr.
For me it looks Jae already implemented sha256 in gno.

https://github.com/gnolang/gno/blob/master/stdlibs/stdlibs.go#L159
https://github.com/gnolang/gno/blob/master/pkgs/gnolang/hash_image.go

What do you think??

[albttx]

@r3v4s you're right, but i believe Gno should match as much as possible go stdlib and so set this function under crypto/sha256.

Maybe a following PR could be to remove this std.Hash

cc: @moul ?

[moul]

In general, let's try being as close as possible; but I'm sure many good reasons will show that we could be different.

Right now I suggest keeping both, the idiomatic way and std.Hash. I don't have enough feedback or feeling yet.

[moul]

@jaekwon could you look at this and #589, please?

[albttx]

And what do you think about adding at the same time sha256.New ?

[jaekwon]

sha256.New is difficult... in stdlibs we want to limit the things to things that can be persisted, and native objects like Go's sha256.New cannot be persisted.

So we would have to write a Gno sha256.Digest shim.
Even with the shim, Go's sha256.digest is not exposed, so we would need to port it to pkgs/crypto/sha256exposed (say).
With the shim and state transfer per operation it could work easily assuming it is easy to copy the .digest in and out, and compared to hash operations this should be fairly quick, so I think this is doable.

But it seems like a lot compared to just exposing some helper methods without .New(), for now.

[jaekwon]

This PR as is with Sha256() seems good though!

[albttx]

rebased and fixed lint

[albttx]

@gnolang/core-tech can this be merge ?

[thehowl]

Merging as this is also pretty much the same code-wise as std.Hash... I did notice, however, that std.Hash and sha256.Sum256 are not the same, because std.Hash trims the result to 20 bytes (from 32).

Judging from the references it is mostly used a few times internally, but probably doesn't have a real purpose in real gno code. It may make sense to remove it. I'll make an issue to discuss that