[Trans] Further comments on RFC 6962

[Phillip Hallam-Baker <[e-mail]>]

Following on from my last one that went out early...


2) The transactions add-chain and add-pre-chain only differ in the type of
the data being exchanged. It would be more general to change the signature
of the request message so that it had an explicit 'data type'

3) Shouldn't it be possible for a service to manage multiple logs? I would
think this particularly important in the case of a retrieve-only log which
is serving multiple logs. But the same point can be made about the add
operation.

4) sha256_root_hash.

Really? This should be an object that has an algorithm/data pair. Encoding
the algorithm into the tag is going to make algorithm agility hard.

5) 'in decimal'

This phrase crops up repeatedly and I don't think it means what is
intended. All JSON integers are encoded in base 10. Using the term 'in
decimal' is confusing as it is distinct from 'as a decimal'

Integer    42       // An integer value in base ten
Decimal   3.14    // A decimal fraction

6) The tree_size of the first tree, in decimal.

In this context, tree_length would probably be easier to see the context.

7) https://<log server>/ct/v1/get-sth-consistency does not specify if the
first and second tree_size identifiers are ordered or not.

It is likely to result in problems if servers assume first > second or
first < second. Either an order should be specified or there should be an
explicit statement that it does not matter.

8) 0-based index

Not clear to me from the reference section if this is a byte count or an
object count. It should not be necessary to hunt through the spec to find
out which.






-- 
Website: http://hallambaker.com/