Suggestion that the plan should be user-focused (deliverable features) rather than specification or implementation details
Rita is available to pair with Pawel to walk through Mercury deliverables. Pawel Szulc (Unlicensed) please coordinate with Rita to find a time for this walkthrough
Łukasz Gołębiewski (Unlicensed) - in RSpace - we have to fix to remove the syncVar if we can live with a mixed store. Keep hot store in memory, concurrent trie map and HistoryTrie in LMDB.
Dom is working on it, should know by the end of this week, benchmark & see if it is correct.
Need to check how it behaves wrt to memory utilization and consumption.
It's the single writer thread - that is causing the bottleneck. Keeping data in memory will help us avoid serialization.
RSpace also needs to use a new threading model.
Planning on this work right away. (this or next sprint)
Medha showed a graph from Jira showing a projection of feature complete in Jan.
This is based off of story point estimates + velocity
January release would be updating of shard parameters + economics finalization.
Requirement to release in Q1 relates to contractual obligations between the Coop and Pithia
Extra-protocol sharding must precede in-protocol (level II) sharding
Question of when which gets delivered when.
We added Peek to Mercury.
Performance of the system
Concerned about the following items relating to performance
Our Grafana dashboards are still not working after the switch of how we report metrics. Medha Parlikar (Unlicensed) confirm this is ticketed so it can be done next sprint.
We don't have baselines yet for performance that are socialized amongst the team
We don't have a person or team in charge of performance. Yet this is the most important deliverable we have.