Cyclic imports are detected by python (they do depend a bit on a sequence of importing stuff).
We do have cyclic dependencies instead that pylint detected, but solving it seems to be a high price to pay for all the worries of new contributors who get some random errors. I was one of the biggest proponents of solving those cyclic dependencies, but looking at some recent problems of various people, I honestly think it's not worth it at the end :). Call it growing-up or maturing. :D. I feel I have grown up a bit by seeing the consequences of it. J. On Wed, Jun 23, 2021 at 8:38 PM Xinbin Huang <bin.huan...@gmail.com> wrote: > > I would like to deprecate it too, so count +1 from me. One question I have > is: do we have any ways to detect and prevent cyclic imports? > > On Wed, Jun 23, 2021 at 11:30 AM Jarek Potiuk <ja...@potiuk.com> wrote: >> >> I think this subject has been raised a few times (last time by Ash). >> Finally I grew up to embrace it as well. >> >> I think I am also fed-up by random pylint errors. Last time >> https://github.com/apache/airflow/pull/15634/checks?check_run_id=2896761068 >> >> We have many, many pylint exceptions all over our code. I can't >> remember the last time where pylint prevented any real error. As Ash >> (rightfully) mentioned in latest discussion on slack - we have >> mypy/flake/isort/black which report (and fix) vast majority of things >> pylint reports. >> >> I think this last error was the final drop for me. >> >> Should we remove pylint ? >> >> Consider it +1 from my side. >> >> J . >> >> >> -- >> +48 660 796 129 -- +48 660 796 129
