OpenClaw: The Great Resource Cleanup
A wave of fixes today centers on one core problem — code that shuts down or times out too eagerly, closing databases, sessions, and plugin resources while work is still in flight. The fix pattern is consistent: keep resources alive until actual work settles, not just until a result is reported.
Duration: PT2M39S
Episode overview
This episode is a short developer briefing from OpenClaw.
It explains recent repository work in plain language.
- Show: OpenClaw
- Published: 2026-09-10T06:05:21Z
- Audio duration: PT2M39S
Transcript excerpt
This excerpt keeps the crawler page concise. Listen to the episode or use the RSS feed for the full update.
Good morning. It's September 10th, and today's OpenClaw activity has a clear throughline: don't close the door while someone's still inside.
Across a dozen-plus pull requests, the team hunted down places where cleanup ran ahead of actual work. Peter Steinberger's PR 143725 releases model-selected plugin resources only after the last agent run truly finishes, not when the first caller reports done. The same logic shows up in PR 143592 and PR 143633, where…
Sessions and state got the same treatment. PR 143602 stops session cleanup from triggering a full disk scan on the main thread when a cached database handle gets evicted. PR 143574 makes sure a rollback doesn't erase the original error message with a generic "database is not open." And a same-day commit, 1c0d32c,…
Second theme: correctness under concurrency and timing. PR 143590 fixes memory search timing out during slow embedding cold starts by pausing the clock correctly during readiness waits. PR 143722 fixes streamed replies splitting mid-line or duplicating code blocks — a subtle bug in how the chunker tracks open code…
Smaller but notable: PR 143519 stops macOS from silently discarding Gateway crash logs before the logger…
What'…