support for Curve25519 in WebCrypto.

[dontcallmedom]

A proposal to add Curve25519 was brought to the TAG:
https://github.com/tQsW/webcrypto-curve25519/blob/master/explainer.md

I thought it would make sense to log it here

[chesnokovilya]

Another important curve specified already in TLS standard https://tools.ietf.org/html/rfc8446 and CFRG https://tools.ietf.org/html/rfc7748 is curve448. I think we need to add it to the web crypto. Here is my quick post in wicg https://discourse.wicg.io/t/proposal-add-curve-25519-and-cruve-448-to-the-web-cryptography-api/4153

[sideshowbarker]

Dom, if any members of the TAG or others would like to see this proposal get some consideration, somebody needs to do the work of engaging with browser-engine implementors about it here.

[dontcallmedom]

@sideshowbarker I belive @tQsW works for an implementer (Google) and the proposal had received contributions from @martinthomson who works for another (Mozilla); paging @wseltzer who was involved in some related discussions I believe

[vlovich]

I wanted to highlight an ambiguity I noticed (tQsW/webcrypto-curve25519#8) when I was reviewing NodeJS's implementation (nodejs/node#36879) which seems to allow import/export of private keys in raw format for Ed25519. Hoping that ambiguity is clarified in any standardization.

[twiss]

There's now a WICG draft for Curve25519 and Curve448 in Web Crypto, here. That's being / been discussed in #196, so I'll close this as a duplicate of that issue.