perf: replace Python delta loop with zstd C extension for bundle compression
compute_delta was pure Python — O(n) hash table per blob, 18s for 5162 real-world objects.
Replaced with zstandard.ZstdCompressor(level=3) — one C call per object. Encoding is 'zstd'; falls back to 'raw' when compression expands the blob.
Wire size reduction: ~4-6x for Python code (149 MB → ~25 MB expected). Build time: < 1s regardless of blob size or count.
sha256:8bdc83c880d8701acd37616d9b21cc53a8758ffa87374643fcd11edac7f0ab74
sha
+3
~3
−3
symbols
sha256:52e17179f1babc186c484afc55c7a5736ea0cb51e1f7225503965060ba6e9b12
snapshot
+3
symbols added
~3
symbols modified
−3
symbols removed
0
dead code introduced
Semantic Changes
9 symbols
~
tests/test_bundle_object_compression.py
.py
3 symbols added, 2 symbols removed, 2 symbols modified
0 comments
To add a comment, use the Muse CLI:
muse hub commit comment sha256:8bdc83c880d8701acd37616d9b21cc53a8758ffa87374643fcd11edac7f0ab74 --body "your comment"
No comments yet. Be the first to start the discussion.