Paul J. Dorn comments

Results 153 comments of


                                            Paul J. Dorn

--limit-request-line check default to MAX_REQUEST_LINE

Linking commit https://github.com/benoitc/gunicorn/commit/d79ff999ce895e2ed0ea02aa8729e6da736dfc27 and comment https://github.com/benoitc/gunicorn/commit/b7b0979ad9fdb53d0a1287f29e78a8daa987831f#r2629719 for easier inquiry into *why* it was swapped around. Imho capping both lower and upper bound are wrong, we should just respect what was...

Sometimes running the fuzzer causes assertion failures

I hit that a lot.. appears recoverable in

CI: minimal dependency versions: specify them and try them in CI

Attached further patches: * CI should verify that whatever changes and workarounds are applied by `debhelper` do not break building * the best way to do that would be to...

add changelog to project_urls

This was applied as 5e30bfa6b1a3e1f2bde7feb514d1734d28f39231

Use plain socket objects instead of wrapper classes

Should the monkey patching in `gunicorn/workers/ggevent.py` stay the same? I [was able](https://github.com/pajod/gunicorn/actions/runs/9153675753/job/25163004748#step:8:118) to run my WIP [request after reloading test](https://github.com/pajod/gunicorn/blob/aa63761ac99654296ee0a3b324f771fa96356945/tests/test_e2e.py#L291) against your PR after fixing just the following variable access:...

--reload not working for docker

uvicorn is deprecating shipping such worker: https://github.com/encode/uvicorn/commit/772c24bd1940fa81ad5f670c46f7d98615079838 and the suggested replacement has the related open issue: https://github.com/Kludex/uvicorn-worker/issues/7

CI: Pylint 3.x

@tilgovi Care to give this another look? Updated a bit after extracting #3131 into a separate PR.

CI: Pylint 3.x

@benoitc 2nd attempt for comment marked "outdated" above. I want newer pylint versions to stop suggesting that inotify might be undefined in the separate conditional, without suppressing that (generally useful!)...

alias CLD to CHLD in signal handlers

Most of them, just not macOS. Python has even recently [updated ](https://github.com/python/cpython/issues/89414) their documentation to clarify that [SIG*](https://docs.python.org/3.13/library/signal.html#signal.SIGCLD) is is just whatever the C library says - hence no such...

Add config: buf_read_size

@aamirawan7584 The way we use it, a larger setting than what was sent should not negatively impact the processing of small requests. However, incorrectly reducing our buffer sizes in response...