-
Notifications
You must be signed in to change notification settings - Fork 27
Issues
is:issue state:open
is:issue state:open
Issue creation is restricted in this repository
Search results
marmot-app: subscribe_timeline_messages / subscribe_chats / subscribe_chat_list / subscribe_group_state run their initial SQLCipher snapshot read synchronously on the caller thread, while sibling subscribe_messages offloads via blocking_app_task (distinct sites; follow-up to #740/#762)
performancePerformance issuePerformance issueStatus: Open.#802 In marmot-protocol/mdk;incident-replay: recover_fork computes
source_epoch + 1unchecked — panics (debug) / silently wraps (release) on an adversarial or corrupt Goggles exportbugSomething isn't workingSomething isn't workingStatus: Open.#801 In marmot-protocol/mdk;marmot-markdown: block-level link-reference-definition label has no length cap, unlike every sibling label scanner — accepts spec-invalid >999-char labels
bugSomething isn't workingSomething isn't workingStatus: Open.#800 In marmot-protocol/mdk;storage-sqlite: one undecodable app_events.tags_json row permanently poisons a group's entire timeline rebuild (no self-heal), unlike the search/unread paths which tolerate it (sibling of #769)
bugSomething isn't workingSomething isn't workingStatus: Open.#799 In marmot-protocol/mdk;transport-quic-broker: publish hot path runs a full O(max_rooms) room sweep + total-backlog recompute under the single global BrokerStateInner mutex on every forwarded record
performancePerformance issuePerformance issueStatus: Open.#797 In marmot-protocol/mdk;transport-nostr-adapter: relay sync-timing telemetry + initial-sync gate key the same relay under two different RelayIndex values (verbatim start endpoint vs RelayUrl-canonicalized inbound endpoint) — EOSE/first-event silently dropped,
subscription_syncednever reaches truebugSomething isn't workingSomething isn't workingperformancePerformance issuePerformance issueStatus: Open.#795 In marmot-protocol/mdk;cgka-engine: direct-ingest inbound commit mirrors the Marmot group record + self-cap cache OUTSIDE the merge transaction — a crash/SQLITE_BUSY between them permanently desyncs the record from MLS state
bugSomething isn't workingSomething isn't workingStatus: Open.#794 In marmot-protocol/mdk;- Status: Open.#792 In marmot-protocol/mdk;
transport-nostr-peeler:
PeeledMessage.group_idis populated with the 32-byte transport routing id wrapped in aGroupId, conflating the two id spaces the invariant separates (latent)bugSomething isn't workingSomething isn't workingStatus: Open.#788 In marmot-protocol/mdk;transport-quic-stream: agent-text-stream record AEAD has no per-session salt — re-publishing to the same anchor within an epoch reuses the ChaCha20-Poly1305 (key, nonce) pair
bugSomething isn't workingSomething isn't workingStatus: Open.#787 In marmot-protocol/mdk;