Adapt code for flake8 >= 3.8.0 #23
Conversation
|
It seems Travis-CI check is broken in this repo (build was triggered, but build status is not shown in the pr). |
|
Ok, only py2 tests are broken now, see the build logs. It seems, this was broken long time ago by newer pytest or hypothesis. Maybe we can require py3? Projects on the Github, using flake8-rst, like pandas, seems to be py3-only. |
* Application.__init__() modification may be drasticaly reduced if the upstream removes the hardcoded arguments for manager.OptionManager. * Reimplemented RstFileChecker.reporter() * Added disable_noqa option with default value for tests Closes #22
For a while, use workaround from flake8-docs/flake8-rst#23. Closes diofant#1052
|
I'd love to see this merged. Just burned half an hour troubleshooting this. :) |
|
@Cabalist, it seems @kataev doesn't care about his pet anymore... Also, the patch seems to be pretty big. Unfortunately, I doubt we can do anything with this: upstream closing even simple merge requests, which address this issue. Apparently, they won't support some stable API for project's like flake8-rst. They suggest instead insane things like parsing the flake8 output... |
|
Oh well, let's go merge this :) |
|
I will make release today, sorry for delay! |
|
@kataev, maybe you can convince flake8-people to accept something like 458 ? (This one - kills allmost all my modifications in the flake8_rst/application.py) I would expect, that the current using flake8 will be quite fragile, unless we find some way to interact with the upstream. Or, if we reimplement flake8-rst, using public interfaces, which they offer... |
This is a quick fix, which address #22. It seems to be working for me (I tried this on the diofant codebase), but some tests are failing.
I hope, @kataev, you will find a time to review this pull request and help with it's finishing. BTW, Application.__init__() modification may be drasticaly reduced if the upstream removes the hardcoded arguments for manager.OptionManager.
Closes #22