Arvo now supports receiving scry requests as HTTP GET requests (as of this PR↗), but they are not optimized in the runtime. What remains to be done is:
- Add dispatch logic to the runtime to submit an Arvo +peek (read) request instead of a +poke (write) request that injects an event, based on whether the path in the URL indicates the request is a scry request, i.e. it starts with /~_~/.
- Ensure the runtime honors whether the request is logged in using the standard auth token. There is prior art for this check, in the runtime's static HTTP response handler.
- Add logic to look up the scry request using the runtime's existing HTTP scry cache.
- Add logic to populate the HTTP scry cache when receiving a scry response from Arvo.