Updated RSA DID multiformat #550
Conversation
| ensureM $ sendRequest IPFS.getPeers | ||
| ensureM $ sendRequest getIPFSPeers |
There was a problem hiding this comment.
That's interesting! Why did you decide to switch the naming convention? I personally like the Something.x convention, so I'm curious :)
There was a problem hiding this comment.
I also prefer that convention. This is from the new Servant.Generic client, since it's easier to decompose them in one file rather than importing the outer level into separate files and exposing them there. The tradeoff is that it puts them in a flat namespace.
| RSAPublicKey rsa -> | ||
| 0x12 : 0x05 : BS.unpack (Lazy.toStrict . Builder.toLazyByteString . getUtf8Builder $ display rsa) |
There was a problem hiding this comment.
Just randomly took a stroll through here, and now I know this is actually incorrect, the prefix needs to be 0x85 : 0x24 : Bs.unpack ...
I've implemented it that way in ts-ucan and it tests fine with the test vectors from the DID spec.
https://github.com/ucan-wg/ts-ucan/pull/42/files#diff-109e954c94a80ae8fd78b2ce9734642de7a787f48926c9a5ba18ad1ec76225ffR17
The reason it's 0x85 0x24 is that that's the number 0x1205 encoded as unsigned varint.
Just like a couple of lines above, the multicodec for ed25519-pub 0xed is encoded as unsigned varint into 0xed 0x01.
There was a problem hiding this comment.
Yup, that's a bug. ALSO the entire format has since been updated in the spec (I assume that ASN.1/DER PR is getting merged), so we need to rewrite a bunch of this PR
| |> X509.pubKeyToPEM | ||
| |> PEM.pemWriteBS | ||
| [X509.PubKeyRSA pk] | ||
| |> X509.writePubKeyFileToMemory |
There was a problem hiding this comment.
These end up larger than before.
writePubKeyFileToMemory = pemsWriteBS . map pubKeyToPEMSoooo, that's in PEM with headers and stuff. This is definitely wrong.
Probably want
PEM.pemWriteLBS
expede
force-pushed
the
ucan-rsa-multiformat
branch
from
e30022c
to
cad0120
Compare
expede
force-pushed
the
ucan-rsa-multiformat
branch
from
877dd7c
to
0e90383
Compare
expede
force-pushed
the
ucan-rsa-multiformat
branch
from
3e0841d
to
f7f16f4
Compare
|
Rebased. Tests failing, and need to validate the format against what Qri has (and then contribute a test vector in the Multicodec repo) |
|
Waiting for the dust to settle over here: multiformats/multicodec#235 |
|
The W3C folks merged the format update. MUAHAHHA this is back on the docket 🚀 (The compact representation was dropped — ASN.1/DER was merged here w3c-ccg/did-method-key#41 including test vectors) |
|
Closing in favor of #573 |
* Write DID signature verification & tests * Write failing rsa DID test * Move over DID tests * Manually import core changes from ##550 * Niv update * Remove redundant test * Parse ASN1 DER RSAPublicKeys * Adjust test fixture to use new format * Fix fission-cli type errors * Get everything compiling * Fix one remaining type error * Fix type error * Revert "Niv update" This reverts commit 3a11d68. * Cherry-pick some changes from "Niv update" * Remove AllowAmbiguousTypes * Move Test.Prelude to Test.Web.UCAN.Prelude * Re-add AllowAmbigousTypes for DelegationSemantics I'm pretty sure I need this? https://github.com/fission-suite/fission/runs/5005311975 * Revert encoding DIDs in new format The rest of the system needs to understand the new format first! Still, we can now *consume* the new format in the server/CLI. * Remove unused orphan
rsa-x509-pubhas made it into Multicodec. This PR updates theExchange test cases with Qri teamTest against W3C vectors