I'll do some test with and without jose tomorrow and report my findings!

Alright! I've built deno from your crypto-ec-import-export branch, @seanwykes, and tested with this code: https://gist.github.com/oles/7b1f4569805a7606056d0c867821d4c7

It no longer errors out with Uncaught (in promise) NotSupportedError: Unrecognized algorithm name, and the imported key looks like this:

CryptoKey {
  type: "private",
  extractable: false,
  algorithm: { name: "ECDSA", hash: undefined },
  usages: [ "sign" ]
}

Which seems promising, but it then errors out at the signing:

error: Uncaught (in promise) TypeError: CryptoKey does not support this operation, its algorithm.namedCurve must be P-256
  return new TypeError(`CryptoKey does not support this operation, its ${prop} must be ${name}`)
         ^
    at unusable (https://deno.land/x/[email protected]/lib/crypto_key.ts:4:10)
    at checkSigCryptoKey (https://deno.land/x/[email protected]/lib/crypto_key.ts:89:38)
    at getCryptoKey (https://deno.land/x/[email protected]/runtime/get_sign_verify_key.ts:8:5)
    at sign (https://deno.land/x/[email protected]/runtime/sign.ts:8:27)
    at FlattenedSign.sign (https://deno.land/x/[email protected]/jws/flattened/sign.ts:134:29)
    at CompactSign.sign (https://deno.land/x/[email protected]/jws/compact/sign.ts:47:39)
    at SignJWT.sign (https://deno.land/x/[email protected]/jwt/sign.ts:54:16)
    at file:///home/oles/projects/deno-test/jwt-test.js:40:8

So I changed the

const cryptoKey = await jose.importPKCS8(privateKey, 'ES256')

to

let cryptoKey = await jose.importPKCS8(privateKey, 'ES256')

cryptoKey.algorithm.namedCurve = 'P-256'

AND IT WORKED 🥳

You shouldn't have to do that. The webcrypto implementation should.

You shouldn't have to do that. The webcrypto implementation should.

As I thought, but glad it could be set manually to verify the rest!

The key should look like this:

CryptoKey {
  type: 'private',
  extractable: false,
  algorithm: { name: 'ECDSA', namedCurve: 'P-256' },
  usages: [ 'sign' ]
}

no hash and a namedCurve.

@oles, @panva - quick fix done.

Current jose status - FAILED. 106 passed; 40 failed; 0 ignored; 0 measured; 0 filtered out (3s)

Going in the right direction.

@panva - I'm having trouble debugging the jose test cases, for example, test-deno/jwe_asymmetric.test.ts.
Using VSCode or cmd-line/Chrome Dev Tools, the tests execute before the debugger attaches.

Have you seen this behaviour before? I'm on a MAC/M1.

@panva - I'm having trouble debugging the jose test cases, for example, test-deno/jwe_asymmetric.test.ts. Using VSCode or cmd-line/Chrome Dev Tools, the tests execute before the debugger attaches.

Have you seen this behaviour before? I'm on a MAC/M1.

@seanwykes try running with --inspect-brk it will break on first line of code and you'll be able to set a breakpoint where you want.

@panva - I'm having trouble debugging the jose test cases, for example, test-deno/jwe_asymmetric.test.ts. Using VSCode or cmd-line/Chrome Dev Tools, the tests execute before the debugger attaches.
Have you seen this behaviour before? I'm on a MAC/M1.

@seanwykes try running with --inspect-brk it will break on first line of code and you'll be able to set a breakpoint where you want.

Done that. it just runs the tests without stopping.

Debugger listening on ws://127.0.0.1:9229/ws/289230fb-d902-4f52-a2e7-6c81c1da6dd2
running 7 tests from file:///Users/sean/Projects/deno/jose/test-deno/jwe_asymmetric.test.ts
test Encrypt/Decrypt ECDH-ES crv: P-256 ... FAILED (57ms)
test Encrypt/Decrypt ECDH-ES crv: P-384 ... FAILED (10ms)
test (expecting failure) Encrypt/Decrypt ECDH-ES crv: P-521 ... ok (6ms)
test Encrypt/Decrypt RSA-OAEP-256 ... ok (92ms)
test Encrypt/Decrypt RSA-OAEP-384 ... ok (236ms)
test Encrypt/Decrypt RSA-OAEP-512 ... ok (195ms)
test Encrypt/Decrypt RSA-OAEP ... ok (189ms)```

It's not expected that this PR makes everything pass. But to answer your question, no i have no idea how to debug a deno runtime.

Done that. it just runs the tests without stopping.

Debugger listening on ws://127.0.0.1:9229/ws/289230fb-d902-4f52-a2e7-6c81c1da6dd2
running 7 tests from file:///Users/sean/Projects/deno/jose/test-deno/jwe_asymmetric.test.ts
test Encrypt/Decrypt ECDH-ES crv: P-256 ... FAILED (57ms)
test Encrypt/Decrypt ECDH-ES crv: P-384 ... FAILED (10ms)
test (expecting failure) Encrypt/Decrypt ECDH-ES crv: P-521 ... ok (6ms)
test Encrypt/Decrypt RSA-OAEP-256 ... ok (92ms)
test Encrypt/Decrypt RSA-OAEP-384 ... ok (236ms)
test Encrypt/Decrypt RSA-OAEP-512 ... ok (195ms)
test Encrypt/Decrypt RSA-OAEP ... ok (189ms)```

Right... This is actually #12584

It's not expected that this PR makes everything pass. But to answer your question, no i have no idea how to debug a deno runtime.

I am interested in two fails ..
test Encrypt/Decrypt ECDH-ES crv: P-256 ... FAILED (57ms)
test Encrypt/Decrypt ECDH-ES crv: P-384 ... FAILED (10ms)

I'll run as isolated code and not tests to see what's up.

It's not expected that this PR makes everything pass. But to answer your question, no i have no idea how to debug a deno runtime.

I am interested in two fails .. test Encrypt/Decrypt ECDH-ES crv: P-256 ... FAILED (57ms) test Encrypt/Decrypt ECDH-ES crv: P-384 ... FAILED (10ms)

I'll run as isolated code and not tests to see what's up.

Ye, the only thing missing in those two tests was jwk export (on encrypt), and import (on decrypt) of the ECDH key in dist/deno/lib/encrypt_key_management.ts:50 and src/lib/decrypt_key_management.ts:48. The algorithm.namedCurve attribute that was missing above in some case is also needed.

PR superseded by #13088 and #13104. Closing.