monero-ccs-proposals/vtnerd-2021-q1.md

layout	title	author	date	amount	milestones	payouts
cp	vtnerd Full-Time 2021 Q1	Lee Clagett (vtnerd)	February 17, 2021	133.4 XMR	name funds done status Month 1 (160 hours) 44 XMR 09 April 2021 finished name funds done status Month 2 (320 hours) 44 XMR 19 May 2023 finished name funds done status Month 3 (480 hours) 45.4 XMR 14 July 2023 finished	date amount 4 January 2023 44 date amount 12 June 2023 44 date amount 26 July 2023 45.4

3-Months full-time software development on monero "core" components in 2021 q1 (through early q2).

What

Work primarily on the monerod, wallet2, and monero-lws (pending feedback in comments). Some of the work to be attempted or investigated:

• Complete work necessary to merge DANE/TLSA in wallet2/epee.
• Adding trust-on-first-use support to wallet2 (already work in progress)
• Progress on my proposal to add encryption to p2p protocol has been slow, but it looks like the consensus is FOR encryption, just no clear consensus on TLS 1.3 vs Noise protocol. Continued work would be design+implementation for the selected encryption technique.
• Potential performance improvements:
  • Improve "cryptonote" read-serialization performance - helps with p2p block responses, RPC responses, among other things. Profiling suggests this could have big improvement on efficiency.
  • Determine a technique for safely returning std::weak_ptr/std::shared_ptr of p2p connections - this can help with efficiency in a few areas:
    • Remove lock acquire to send dandelion++ fluffs and I2P/Tor white noise
    • Remove lock acquire and atomic increment per connection when relaying bocks (a list is already obtained)
    • Others? Any foreach_connection call that can be safely done concurrently can use this technique
  • Various epee/p2p serialization (output) routines (carryover from q4, this is difficult to update)
• Potential compilation improvements (carryover from q4 since mj-xmr has focused on this more):
  • Moving serialization code to cpp (although this could be spurious in updates which is frowned upon)
  • Dropping the templated nature of the epee TCP server would yield best improvement, but might be cause too much "pain" in the review
• monero-lws work (assuming no complaints in comments):
  • Provide pre-built binaries
  • Add support for immediate mempool scanning (via ZMQ-PUB)
  • (Unlikely this quarter) - reproducible builds so community members can verify+sign the binary hashes

More than 3 months listed there, priorities may change as community requests or issues arise.

Who

Lee Clagett (vtnerd). I've had two CCS proposals. The full list of my PRs for last quarter CCS. There's also some investigation/writeup for encrypted p2p, and investigations to improve efficiency (logging in particular) that aren't in that list since any code is still a crude mashup for quick testing. A working, but still in progress DANE/TLSA wallet implementation also exists.

Bigger higlights: improved serialization performance to epee binary, removed copy of all outgoing p2p messages and duplicate copies of p2p block notify messages (not merged), changed p2p connection list to std::weak_ptr for small efficiency improvement and code clarity (not merged), improved byte_stream alllocation efficiency, added byte_stream realloc to reduce memory usage after new reallocation strategy (not merged), and fixed timeout bug in Dandelion++.

Proposal

Work on the various tasks outlined above for 40 hours/week over the next 3 months after potential funding. I already use time-tracking software for work; if the hours dip in a given month unexpectedly, the update/milestone will be at the completion of the hours listed above.

The funds were calculated with 55 USD/hour with ~197.85 USD/XMR which is the 14-day exponential moving average on Kraken through 2021/02/16.