Draft local state query serialization tests #1501
Conversation
This is not working as the decoder used is not including the N2C local state query message tag (3), while the api example is using the full message as sent by the client.
This is to use the clientCodecs from the network package.
This is using the socket/plug definition to compose a cddl that actually verifies the whole query message
Concatenating is fine with simple examples, however I really doubt one can define the block CDDL keeping their sanity with only concat. I do think
I'm preparing a validator here: input-output-hk/cuddle#68. As opposed to
No objections from my side. One thing that I would like to note here is that GetSystemStart is presumably one of the easiest specs. I want to warn not to commit to something that works for this simple case and then becomes a horrible nightmare to adapt it to the multi thousand lines CDDLs for blocks and others. Note this suggestion is both for these ouroboros-consensus stuff as well as the CDDLs in the blueprint itself. |
Experimented with how we would want to test against CDDLs in https://github.com/cardano-scaling/cardano-blueprint/ from this repository.
This is incomplete, but I'd like to gather some feedback already on:
UTCTimenot fully roundtripping becausedayOfYearis clamped to[1, 366]- would we want that anyCBORthat is valid against aCDDLto roundtrip, i.e.CBOR -> result -> CBOR?.cddlfiles, while @jasagredo started using cddl module extensions. However that requirescddlcand I was not feeling like packaging that for NixOS. Do we really want/need;# include? If yes, is thecddlcgem the way to go or should we extend https://github.com/input-output-hk/cuddle?CBORagainstCDDLcan only be done with thecddlgem, which is not open source (anymore?) and it's error output is just insane (thousands of repeated lines of errors). Do we want Add support for validating CBOR input-output-hk/cuddle#7 instead?cardano-blueprint?