Just a short progress info:
Improved stability a lot. Spent hours with Valgrind and American Fuzzy Lop and fixed all what has been found so far. Afterwards a complete AFL cycle performed without finding any new issues.
CommonMark 0.27 test suite:
- 610 tests pass (~98%)
- 12 tests fail (~2%)
Cmark’s make bench:
- CMark: mean = 0.6590, median = 0.6600, stdev = 0.0032
- MD4C: mean = 0.4410, median = 0.4400, stdev = 0.0032 (~ 33.1% faster)
Cmark’s pathological_tests.py:
- Cmark: 0m0.803s
- MD4C: 0m0.517s (~ 35.6% faster)
(Tested with current git heads of MD4C and Cmark on a Linux box)