[#A] more semantic types

Can we piggyback on JSON-LD/JSON Schema IRIs to write down semantics of fields? http://www.w3.org/TR/json-ld/#the-context

unix timestamp, tai64na

url

or IRI, https://en.wikipedia.org/wiki/Internationalized_resource_identifier)

scavenge msgpack, bson, postgres, mysql, sqlite etc for what types they’ve ended up defining

website: talk about rpc

specifically, layering as HTTP/2 POSTs, bidir assumption

sequential vs concurrent protocols

sequential: can switch protocols
concurrent: likely faster, goroutine started for each []byte
mention msgp’s rpc lib

better split the spec into normative and non-normative?

push things like arrays of variable size items into a “conventions” section

clearer optional features for library implementations

[#B] website: write a boltdb example

[#B] chitin schema file formatter

be like gofmt
use tabs

write advocacy: no wire overhead for putting a Message in a sized container

e.g. in a key-value database
that is, no field types etc
not counting field lengths as overhead, those are in every format