[MPC] Phase2 MPC

[dtebbs]

Functions and cli utils to perform the phase2 MPC.
Some things that are still WIP / need discussion:

• sources of randomness
• read/write compressed points
• hashing to G2
• robust file operations

(depends on #68)

[AntoineRondelet]

Please add some comments to clarify some of the choices you took in the chacha files, or refer to the corresponding sections of the implementation you followed

[dtebbs]

template<>

Let's either add a comment if necessary rather than using an empty template. Templatized code aims to be written in a tcc file anyway.

(@AntoineRondelet for some reason, github is not letting me reply directly, so I've had to move this to a new comment).

I've added some comments to explain these functions and why they are here. Just in case it's not clear, this code will only work for the case where ppT == libff::alt_bn128_pp, and the empty template here is to specify that this is a concrete function for the case ppT == libff::alt_bn128_pp, rather than generic templated code.

Since this is a concrete function, if it was in a .tcc file it would likely cause link errors due to the symbol being defined multiple times. If we turned it into fully generic code with template<typename ppT> then it would not compile for any other ppT, and it would make it difficult to provide implementations for other curves in the future.

Let me know if the comments are not clear, or you think we should take a different approach.

[AntoineRondelet]

template<>

Let's either add a comment if necessary rather than using an empty template. Templatized code aims to be written in a tcc file anyway.

(@AntoineRondelet for some reason, github is not letting me reply directly, so I've had to move this to a new comment).

I've added some comments to explain these functions and why they are here. Just in case it's not clear, this code will only work for the case where ppT == libff::alt_bn128_pp, and the empty template here is to specify that this is a concrete function for the case ppT == libff::alt_bn128_pp, rather than generic templated code.

Since this is a concrete function, if it was in a .tcc file it would likely cause link errors due to the symbol being defined multiple times. If we turned it into fully generic code with template<typename ppT> then it would not compile for any other ppT, and it would make it difficult to provide implementations for other curves in the future.

Let me know if the comments are not clear, or you think we should take a different approach.

Yes. What I was saying was basically: if this is not generic code, then let's remove the template<> line, and add a comment saying "this code will only work for the case where ppT == libff::alt_bn128_pp"

[EDIT] I thought the compiler would not complain if we removed template<> as the type was specified, but apparently it does, so let's keep the template<> here then

[AntoineRondelet]

LGTM