Wasthon — tests & benches

Each of Wasthon's 25 ported modules has a test-*.html (smoke correctness) and a bench-*.html (perf vs Brython equivalent), with two grouped exceptions: hashlib bundles md5/sha1/sha2/sha3/blake2/hmac, cluster bundles _random/_struct/unicodedata.

▶ Run all tests — sequential runner across every test-*.html.
▶ CPython suite dashboard — runs CPython's own Lib/test/test_*.py against the wasthon modules, with per-module scores (the authoritative behavioural reference).

Bundled targets (one .mjs/.wasm for the whole set)

• test-wasthon (light, 22 modules, ~1 MB — default)
• test-wasthon-full (full, 25 modules, ~3 MB — adds unicodedata + _zstd + _sqlite3)
• test-debug (296 probing tests on the full bundle — fishing for latent bugs)

Tests (smoke correctness)

• array
• binascii
• _bz2
• cluster (_random/_struct/unicodedata)
• cmath
• _csv
• _decimal
• hashlib (md5/sha1/sha2/sha3/blake2/hmac)
• _json
• _lzma
• math
• pyexpat
• _statistics
• _zlib
• _zstd
• _pickle (no bench yet)
• _sre (no bench yet)
• _sqlite3 (:memory: only, no bench yet)

Benches (Wasthon vs Brython)

• array
• binascii
• _bz2
• cluster (_random/_struct/unicodedata)
• cmath
• _csv
• _decimal
• hashlib (md5/sha1/sha2/sha3/blake2/hmac)
• _json
• _lzma
• math
• pyexpat
• _statistics
• _zlib
• _zstd

CPython suite (WIP — behavioural ground truth)

• dashboard (all modules, scored)
• test_struct
• test_math
• test_cmath
• test_array
• test_binascii
• test_csv
• test_decimal
• test_random
• test_hmac
• test_hashlib
• test_zlib
• test_bz2
• test_lzma
• test_unicodedata
• test_pyexpat
• test_pickle
• test_re
• test_sqlite3
• test_statistics
• test_zstd

Diagnostics (dev — memory & bridge internals)

• test-tp-dealloc (A/B proof: refcount→0 dispatches tp_dealloc; toggles runtime.noFree)