search

castedo / dsi-spec

Working Copy of Document Succession Identifier Specification
https://castedo.github.io/dsi-spec/
0 stars 0 forks source link

specification of base DSI hash is weak #1

Open castedo opened 3 months ago

castedo commented 3 months ago

The formal definition of edition 2.3 of the DSI spec (1wFGhvmv8XZfPx0O5Hya2e9AyXo/2.3) reads:

"The base DSI is a base64url representation of a 20-byte hash that identifies a data structure."

As @khinson points out in feedback, it's a rather weak statement, and I agree.

Some thoughts on improvement to follow below.

castedo commented 3 months ago

I'm thinking that until there is another format other than DSGL, the formal section should keep it simple and just say the base DSI is a hash from DSGL. Then in the Discussion section I can go into how formats other than DSGL could be used in the future as long as they have some of the key properties of DSGL:

  1. cryptographically adequate hash of an initial record (adequate for publicly archived scholarly communication)
  2. initial record must identify cryptographic keys of authors of the document succession
  3. only the key holders can amend a document succession
  4. initial record format must accommodate default usage with frequently changing hashes (e.g. every second), as is the case in DSGL

Maybe I should add some material to the DSGL spec mentioning how a datetime is in the init commit and thus a diff DSI is created every second (for the same set of signing keys).