ChainAgnostic / varsig

The cryptographic signature multifomat
Other
11 stars 4 forks source link

💡 Consider other more flexible representations #8

Open Gozala opened 1 year ago

Gozala commented 1 year ago

Given how many times I've thought "wait would different order of parameters be better ?" I started to wonder if perhaps there are other more flexible (order independent) yet compact representations could be used to encode this ?

Couple of options that come to mind:

  1. It may be worth assessing what the overhead would be for dag-cbor if we used single letter field names
    • It would allow extensions in the future
    • It would allow user space extensions as well.
  2. I have no direct experience with borsh but it seems to look good no various benchmarks
  3. I have some good experience with flatbuffers although they're not native to IPLD and I'm not sure how compact tables are in comparison (they are extendable though).
  4. There's also ASN.1 which are used in pkcs8 and spki

I have also though of some multiformats derivative with just a bit more structure so it's bit more than varints which starts to smell like ASN.1.

P.S: I'm not implying we should do this, it just had been on my mind so I thought I'd start a discussion