Amaury Chamayou

Ok, so it seems that: ``` /\ log = ( NodeOne :> 2, contentType |-> TypeReconfiguration, value |-> {NodeTwo}], [term |-> 2, contentType |-> TypeSignature, value |-> 0], [term |->...

> I think at the heart of this discussion is the definition of committable. My understanding is that a committable transaction (one that is followed by a signed transaction) is...

@jumaffre good point, but I think it would still be good to have a generic, suite-able version (something that did a loggingcpp -> loggingjs recovery, or the other way around...

@wintersteiger there are two sub-tests here, the first is historical queries themselves (ie. querying table data across a service identity change). That should work today (there's evidence of users doing...

I'm very keen to try the basic approach of chunking the snapshot and sending it over the ring buffer, even if it means increasing the max message size as well...

@jeffa5 there isn't exactly something like this now, but the closest thing we have, and what we've discussed extending in that direction in the past is the indexing API: https://microsoft.github.io/CCF/main/architecture/indexing.html...

Not sure about "trusted_authority" as a name for these members, it's very generic. In some sense every member is a TA, since we accept signed statements from them. I think...

@jeffa5 sounds reasonable to me, is it fair to characterize the overhead for users who only have `handle_committed_transaction` as minimal?

#4343

We want to default to the public AMD endpoint, because it will work anywhere, and allow the Azure endpoint to be configured where appropriate.