perf: O(N) topological sort + bulk snapshot INSERT in wire_push
_topological_sort used queue.pop(0) on a list — O(N) per pop, O(N²) total. Replaced with collections.deque.popleft() → O(N).
bulk_upsert_snapshot_entries was re-processing existing snapshots on every push (DELETE + re-INSERT entries for content-addressed data that never changes). Now skips snapshots that already exist on the server. Also replaces N session.add() calls with a single Core insert() bulk statement.
Result: 51-commit push_pack 0.946s → 0.139s (6.8x faster).
sha256:bdd4e443023e40c96785778cad2b6947a5ec7a3fb0e628cfc05badfbb5a76082
sha
sha256:b693ea4f87d3b791603a7d0a51130aa58246fc9a913cf995f304d13a0476974a
snapshot
0 comments
To add a comment, use the Muse CLI:
muse hub commit comment sha256:bdd4e443023e40c96785778cad2b6947a5ec7a3fb0e628cfc05badfbb5a76082 --body "your comment"
No comments yet. Be the first to start the discussion.