Claude 2762553f78 perf(desktop): batch bridge replies, virtual-thread router, one trim job ...

- Replies to the page queue and go out together in one script, in
  order: a burst of 100 reads now takes about 60 renderer round trips
  instead of 100. A test checks every reply in a burst reaches its
  request.
- The SOCKS router runs each connection's two blocking pumps on
  virtual threads instead of holding two platform threads each.
- Closing windows asks a single conflated job on the runtime to trim
  the blob cache instead of starting a thread per window.
- The site card parses its paths once per event, and the engine
  reports download progress in whole percents, so it recomposes
  about 100 times per download instead of once per chunk.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01NQ76C3uhMzzWRfB6NNxYzf

2026-10-29 23:00:20 +00:00

..

plans

build: build against quartz 2.6.0

2026-10-28 14:22:45 +00:00

src

perf(desktop): batch bridge replies, virtual-thread router, one trim job

2026-10-29 23:00:20 +00:00

.gitignore

Starting a new module for app benchmarks

2024-02-05 18:18:33 -05:00

ARCHITECTURE.md

refactor(cordn): coordinatorRelayPool over Quartz's INostrClient.cvmRelayPool

2026-10-28 12:23:35 +00:00

build.gradle.kts

build: run :commons:jvmTest in two forks

2026-10-24 02:19:13 +00:00

consumer-rules.pro

Starting a new module for app benchmarks

2024-02-05 18:18:33 -05:00