masklinn

All benching done with samples/useragents.txt (75158 lines, 20322 unique). NOTE: "legacy" has a clearing cache of size 200. |parser|cache (n=200)|cpython 3.12|pypy 7.3.17|graalpy 24.1.0| |-|-|-:|-:|-:| |legacy |clearing | 29.00s (386us/line)| 156.56s...

- https://github.com/oracle/graalpython/issues/445 - https://github.com/pypy/pypy/issues/1347

Reported the issues upstream (see previous comment) and implemented some rewriting on the Python side (#230 only graal and re2 as idk if pypy uses a DFA and uap-rs does...

> Thanks, note that `pytest` step also needs adjustment because you can't run tests for x86_64 wheels on aarch64 macOS after this change. Ah good catch, I assumed either arm64...

> That may not work with pypy I think, they don't have universal2 binaries: https://pypy.org/download.html So should probably pass the architecture to `setup-python` still (same as Windows) in order to...

> Right, but note that it's only for Python >= 3.11. Sure but generate-ci creates a template for 3.x so that's been >= 3.11 since it was created. Though if...

Well apparently it doesn't, python < 3.11 x64 seems broken on M1: https://github.com/masklinn/ci-test/actions/runs/11296863554/job/31422564472 I'll have to see with the setup-python people if they know why. edit: opened actions/setup-python#960

Welp, apparently action/setup-python has decided not to fix the non-universal python images, so not quite sure what to do there, I guess I'll close this.?

Ah I see, I didn't know graal used a dfa internally, that explains things. I've had [the same issue in rust as `regex` is also automata-based](https://github.com/rust-lang/regex/discussions/1206#discussioncomment-9976284), and the way the...

> Currently, no. Our regex engine seems to have a property for that, but we currently don't expose it on the Python Pattern object. (`_sre.tregex_compile(pattern, _sre._METHOD_SEARCH, False).isBacktracking` works, but it's...