On Thu, 16 Mar 2017 13:45:28 +0100 Linus Nordberg <[email protected]> wrote:
> Andrew Ayer <[email protected]> wrote > Wed, 15 Mar 2017 14:48:32 -0700: > > > On Wed, 15 Mar 2017 21:44:46 +0100 > > Linus Nordberg <[email protected]> wrote: > > > >> The gossip protocol should work for both CT v1 and CT v2. If it > >> doesn't, we should fix that. If that's not possible, let's define a > >> gossip protocol version two. > > > > The sth-pollination protocol defined in draft-ietf-trans-gossip-04 > > could work with v1 STHs, but section 8.2.4 says it contains an > > array of v2 STHs: > > > > "sths - an array of 0 or more fresh SignedTreeHeads as defined in > > [RFC-6962-BIS-09] Section 3.6.1." > > Hmm. It seems like CT v1 has been ignored in the transition to > 6962-bis. It seems very desirable to add explicit CT v1 support, as it will probably be quite a while before CT v2 fully replaces CT v1, and there is a need for gossip in the v1 ecosystem in the meantime. I don't think the changes would be significant, and I'd be willing to make a pass through the draft to add text where necessary. However, the gossip draft has already completed WGLC. What kind of changes can be made at this point? > > For this reason, I've been implementing draft-ietf-trans-gossip-00, > > which uses v1 STHs and uses the > > URL .well-known/ct/v1/sth-pollination. > > > > Should I be using the URL defined in -04 instead? > > > > Incidentally, -04 is not entirely clear how STHs are represented. > > RFC6962-bis no longer defines a JSON representation for STHs. > > Instead STHs are returned in JSON responses as base64-encoded > > SignedTreeHeads. Does this mean that the sth-pollination protocol > > should use a JSON array of strings, possibly mixed with JSON > > objects for v1 STHs? > > I don't know right now. Suggestions welcome! Well, I guess your > question is a suggestion. Analysis welcome, as well as proposed text > of course. :) Yes, you can take my question as a suggestion. The advantage is that it avoids creating redundant formats for STHs and allows code reuse for parsing STHs. The disadvantage is that mixing objects and strings in the same array might make parsing the JSON awkward with some libraries/languages. > Also, very happy to see implementation under way! I should have a report to make to trans soon :-) Regards, Andrew _______________________________________________ Trans mailing list [email protected] https://www.ietf.org/mailman/listinfo/trans
