Just to give an example of the problems we have to deal with re: false positives.
I ALMOST had a working Python 3.9 change this morning. ALMOST because when I run it everything worked except about hundred new 'no-member' errors in a file that I have not touched (and it has no relation whatsoever with the changes I made). No new pylint was released. Just other, unrelated code was changed: https://github.com/potiuk/airflow/runs/2903575190?check_suite_focus=true#step:9:48 This is something that recently started to occur almost daily - for various people. This is mightily annoying and causes even more delays and reruns of the builds (thus money) for really bogus reasons. J. On Thu, Jun 24, 2021 at 11:33 AM Ash Berlin-Taylor <a...@apache.org> wrote: > > *23m23s on our more-powerfull self-hosted runners. > > On Thu, Jun 24 2021 at 10:32:43 +0100, Ash Berlin-Taylor <a...@apache.org> > wrote: > > The biggest advantage to me: reducing test time on CI and pre-commit. > > Pylint is slow. > > Oh my god, I just checked and it's even slower than I realised > > 46m25s on Github Runners, 23m23s for the "pylint" job. > > https://github.com/apache/airflow/runs/2902755059?check_suite_focus=true > https://github.com/apache/airflow/runs/2900159473?check_suite_focus=true > > There is no way pylint is worth that much test time! > > -ash > > On Thu, Jun 24 2021 at 11:22:53 +0200, Jarek Potiuk <ja...@potiuk.com> wrote: > > Good point Ash - I think we could indeed (if we decide to disable pylint) do > this very thing indeed - identify the "important" aspects and try to find a > plugin in flake8 for it. With our codebase, I think it will be rather easy to > find if it works well for us (and we could do it as part of the > `pylint-disable` PR and iterate until we are happy). I think it would > streamline the development of ours a bit. J. On Thu, Jun 24, 2021 at 11:12 AM > Ash Berlin-Taylor <a...@apache.org> wrote: > > Some of the "missing" checks can be added by flake8 plugins -- now obviously > the plugins don't always provide the "no-false-positives" guarantees > https://github.com/PyCQA/flake8-bugbear is a good one, and is part of the > PyCQA so is "blessed". There are many more possible ones we can look at on > https://github.com/DmytroLitvinov/awesome-flake8-extensions, but I think we > should keep the plugins small and only add specific plugins when we identify > a need -- to keep flake8 running fast. -ash On Thu, Jun 24 2021 at 10:59:58 > +0200, Jarek Potiuk <ja...@potiuk.com> wrote: Added counts to get better > overview: grep -R '# pylint: disable' | sed 's/.*\(# pylint.*\)/\1/' | sort | > uniq -c 1 Binary file > .git/objects/pack/pack-b85cbf9c91a5ba2bc0e0f19ea9bcf9d8193c48d5.pack matches > 3 # pylint: disable=abstract-method 18 # pylint: > disable=assignment-from-no-return 22 # pylint: > disable=attribute-defined-outside-init 195 # pylint: disable=broad-except 2 # > pylint: disable=C0103 2 # pylint: disable=C0103, line-too-long 1 # pylint: > disable=C0111 2 # pylint: disable=C0123 4 # pylint: disable=C0301 3 # pylint: > disable=C0302 18 # pylint: disable=c-extension-no-member 7 # pylint: > disable=c-extension-no-member, no-member 2 # pylint: > disable=c-extension-no-member,no-member 19 # pylint: > disable=comparison-with-callable 1 # pylint: disable=comparison-with-itself 4 > # pylint: disable=consider-using-enumerate 2 # pylint: > disable=consider-using-generator 46 # pylint: disable=consider-using-with 2 # > pylint: disable=consider-using-with,attribute-defined-outside-init 4 # > pylint: disable=cyclic-import 2 # pylint: disable=dangerous-default-value 2 # > pylint: disable=E0012 1 # pylint: disable=E1101 2 # pylint: disable=E1102 3 # > pylint: disable=E1111 8 # pylint: disable=expression-not-assigned 29 # > pylint: disable=global-statement 1 # pylint: > disable=global-statement,global-variable-not-assigned 4 # pylint: > disable=inconsistent-return-statements 1 # pylint: > disable=invalid-bool-returned, bad-option-value 1 # pylint: > disable=invalid-length-returned 123 # pylint: disable=invalid-name 2 # > pylint: disable=invalid-name,missing-docstring 2 # pylint: > disable=invalid-name, unused-argument 1 # pylint: > disable=invalid-sequence-index 2 # pylint: disable=invalid-str-returned 2 # > pylint: disable=len-as-condition 20 # pylint: disable=line-too-long 3 # > pylint: disable=line-too-long # noqa 4 # pylint: disable=logging-not-lazy 4 # > pylint: disable=lost-exception 26 # pylint: disable=maybe-no-member 21 # > pylint: disable=missing-docstring 1 # pylint: disable=missing-docstring, > redefined-outer-name 12 # pylint: disable=missing-function-docstring 6 # > pylint: disable=missing-kwoa 4 # pylint: disable-msg=too-many-arguments 2 # > pylint: disable=no-else-break 2 # pylint: disable=no-else-continue 741 # > pylint: disable=no-member 1 # pylint: disable=no-member,invalid-name 3 # > pylint: disable=no-member,line-too-long 1 # pylint: disable=no-member, > maybe-no-member 3 # pylint: disable=no-member # noqa 2 # pylint: > disable=no-method-argument 58 # pylint: disable=no-name-in-module 1 # pylint: > disable=no-name-in-module,wrong-import-order 2 # pylint: > disable=non-parent-init-called 6 # pylint: disable=no-self-argument 1 # > pylint: disable=not-a-mapping 6 # pylint: disable=not-callable 222 # pylint: > disable=no-value-for-parameter 6 # pylint: disable=pointless-statement 123 # > pylint: disable=protected-access 2 # pylint: > disable=protected-access,c-extension-no-member,no-member 2 # pylint: > disable=protected-access,no-member 2 # pylint: disable=R0401 2 # pylint: > disable=R0902,R0904 1 # pylint: disable=R0904, C0111 1 # pylint: > disable=R0904, C0111, C0302 3 # pylint: disable=R0904, C0302 3 # pylint: > disable=R0913, C0302 2 # pylint: disable=raising-format-tuple 14 # pylint: > disable=redefined-builtin 1 # pylint: > disable=redefined-builtin,unused-argument 2 # pylint: > disable=redefined-builtin,unused-argument,too-many-arguments 14 # pylint: > disable=redefined-outer-name 6 # pylint: > disable=redefined-outer-name,reimported 7 # pylint: > disable=redefined-outer-name,reimported,unused-import 1 # pylint: > disable=redefined-outer-name,unused-argument 2 # pylint: > disable=reimported,unused-import,redefined-outer-name 2 # pylint: > disable=self-assigning-variable 13 # pylint: disable=signature-differs 2 # > pylint: disable=signature-differs # noqa: D403 2 # pylint: > disable=singleton-comparison 4 # pylint: disable=subprocess-popen-preexec-fn > 2 # pylint: disable=subprocess-popen-preexec-fn,consider-using-with 5 # > pylint: disable=subprocess-run-check 7 # pylint: > disable=super-init-not-called 2 # pylint: disable=syntax-error 3 # pylint: > disable=too-few-public-methods 17 # pylint: disable=too-many-ancestors 270 # > pylint: disable=too-many-arguments 2 # pylint: > disable=too-many-arguments,inconsistent-return-statements 6 # pylint: > disable=too-many-arguments, too-many-locals 26 # pylint: > disable=too-many-arguments,too-many-locals 2 # pylint: > disable=too-many-arguments,too-many-locals,too-many-branches 2 # pylint: > disable=too-many-arguments,too-many-locals, too-many-statements 2 # pylint: > disable=too-many-boolean-expressions 4 # pylint: disable=too-many-branches 1 > # pylint: disable=too-many-function-args 78 # pylint: > disable=too-many-instance-attributes 2 # pylint: > disable=too-many-instance-attributes, too-many-arguments 2 # pylint: > disable=too-many-instance-attributes,too-many-arguments,too-many-branches 2 # > pylint: disable=too-many-instance-attributes,too-many-locals 2 # pylint: > disable=too-many-instance-attributes,too-many-public-methods 14 # pylint: > disable=too-many-lines 21 # pylint: disable=too-many-locals 6 # pylint: > disable=too-many-locals,too-many-arguments 3 # pylint: > disable=too-many-locals,too-many-arguments,invalid-name 3 # pylint: > disable=too-many-locals,too-many-arguments, too-many-branches 4 # pylint: > disable=too-many-locals,too-many-statements 71 # pylint: > disable=too-many-nested-blocks 9 # pylint: disable=too-many-public-methods 18 > # pylint: disable=too-many-return-statements 7 # pylint: > disable=too-many-statements 5 # pylint: disable=undefined-variable 11 # > pylint: disable=unexpected-keyword-arg 4 # pylint: disable=ungrouped-imports > 4 # pylint: disable=unidiomatic-typecheck 1 # pylint: > disable=unnecessary-lambda 18 # pylint: disable=unsubscriptable-object 4 # > pylint: disable=unsupported-membership-test 1 # pylint: disable = > unused-argument 188 # pylint: disable=unused-argument 1 # pylint: > disable=unused-argument,invalid-name 1 # pylint: > disable=unused-argument,line-too-long 2 # pylint: disable=unused-argument # > pragma: no cover 3 # pylint: > disable=unused-argument,too-many-arguments,too-many-locals 2 # pylint: > disable=unused-argument, unused-variable 549 # pylint: disable=unused-import > 2 # pylint: disable=unused-import,line-too-long 2 # pylint: > disable=unused-import # noqa 2 # pylint: disable=unused-import # noqa: F401 4 > # pylint: disable=unused-import # noqa: F401 8 # pylint: > disable=unused-variable 1 # pylint: disable=unused-variable # noqa 4 # > pylint: disable=using-constant-test 1 # pylint: disable=W0104 6 # pylint: > disable=W0106 1 # pylint: disable=W0108 1 # pylint: disable=W0143 2 # pylint: > disable=W0201 20 # pylint: disable=W0212 1 # pylint: disable=W0612 4 # > pylint: disable=W0703 1 # pylint: disable=wildcard-import 5 # pylint: > disable=wrong-import-order 5 # pylint: disable=wrong-import-position On Thu, > Jun 24, 2021 at 10:53 AM Jarek Potiuk <ja...@potiuk.com> wrote: It could be > survivor bias, but at least in my memory pretty much every time I've seen > something really serious, it's been also reported by flake8 (which on its own > is a combination of pep8, pyflakes + circular complexity). Let me explain a > bit more context that built up in my mind over the last few months which made > me change my mind. I think for such a tool, it is important to have as few > false-positives as possible. If you have a lot of false-positives, we are > basically trading both - contributor's and reviewers time on analysing and > explaining that "this is false positive" with the potential of introducing a > mistake like a typo. As a reviewer, this is a recurring pattern recently that > the contributor complains about a "new" problem reported in the code they did > not touch and then the reviewer has to get engaged and explain, and iterate > often resulting in having to rebase the PR (sometimes again and again). When > I look at our code: grep -R '# pylint: disable' | wc -l 3370 find . -name > '*.py' | wc -l 4924 We have > 3000 false positives in our code and almost > 5000 python files. Those false positives are time lost by both contributors > and reviewers. Those are the places where contributor and reviewer jointly > decided that it's not worth to fix the problem reported by pylunt. Is > 3370/4920 a lot ? I think so. Yes it might mean we should disable some rules > possibly. But when I look at those rules to disable, those are the very ones > that we would like to be able to discover. The list of those unique disables > at the end of the mail. If you look at the pyflakes (which is one of the > three tools in flake8) -> "Pyflakes makes a simple promise: it will never > complain about style, and it will try very, very hard to never emit false > positives." . I think it delivers. Pylint makes no such promise (and delivers > as well). J. #### Currently disabled pylint rules: grep -R '# pylint: > disable' | sed 's/.*\(# pylint.*\)/\1/' | sort | uniq # pylint: > disable=abstract-method # pylint: disable=assignment-from-no-return # pylint: > disable=attribute-defined-outside-init # pylint: disable=broad-except # > pylint: disable=C0103 # pylint: disable=C0103, line-too-long # pylint: > disable=C0111 # pylint: disable=C0123 # pylint: disable=C0301 # pylint: > disable=C0302 # pylint: disable=c-extension-no-member # pylint: > disable=c-extension-no-member, no-member # pylint: > disable=c-extension-no-member,no-member # pylint: > disable=comparison-with-callable # pylint: disable=comparison-with-itself # > pylint: disable=consider-using-enumerate # pylint: > disable=consider-using-generator # pylint: disable=consider-using-with # > pylint: disable=consider-using-with,attribute-defined-outside-init # pylint: > disable=cyclic-import # pylint: disable=dangerous-default-value # pylint: > disable=E0012 # pylint: disable=E1101 # pylint: disable=E1102 # pylint: > disable=E1111 # pylint: disable=expression-not-assigned # pylint: > disable=global-statement # pylint: > disable=global-statement,global-variable-not-assigned # pylint: > disable=inconsistent-return-statements # pylint: > disable=invalid-bool-returned, bad-option-value # pylint: > disable=invalid-length-returned # pylint: disable=invalid-name # pylint: > disable=invalid-name,missing-docstring # pylint: disable=invalid-name, > unused-argument # pylint: disable=invalid-sequence-index # pylint: > disable=invalid-str-returned # pylint: disable=len-as-condition # pylint: > disable=line-too-long # pylint: disable=line-too-long # noqa # pylint: > disable=logging-not-lazy # pylint: disable=lost-exception # pylint: > disable=maybe-no-member # pylint: disable=missing-docstring # pylint: > disable=missing-docstring, redefined-outer-name # pylint: > disable=missing-function-docstring # pylint: disable=missing-kwoa # pylint: > disable-msg=too-many-arguments # pylint: disable=no-else-break # pylint: > disable=no-else-continue # pylint: disable=no-member # pylint: > disable=no-member,invalid-name # pylint: disable=no-member,line-too-long # > pylint: disable=no-member, maybe-no-member # pylint: disable=no-member # noqa > # pylint: disable=no-method-argument # pylint: disable=no-name-in-module # > pylint: disable=no-name-in-module,wrong-import-order # pylint: > disable=non-parent-init-called # pylint: disable=no-self-argument # pylint: > disable=not-a-mapping # pylint: disable=not-callable # pylint: > disable=no-value-for-parameter # pylint: disable=pointless-statement # > pylint: disable=protected-access # pylint: > disable=protected-access,c-extension-no-member,no-member # pylint: > disable=protected-access,no-member # pylint: disable=R0401 # pylint: > disable=R0902,R0904 # pylint: disable=R0904, C0111 # pylint: disable=R0904, > C0111, C0302 # pylint: disable=R0904, C0302 # pylint: disable=R0913, C0302 # > pylint: disable=raising-format-tuple # pylint: disable=redefined-builtin # > pylint: disable=redefined-builtin,unused-argument # pylint: > disable=redefined-builtin,unused-argument,too-many-arguments # pylint: > disable=redefined-outer-name # pylint: > disable=redefined-outer-name,reimported # pylint: > disable=redefined-outer-name,reimported,unused-import # pylint: > disable=redefined-outer-name,unused-argument # pylint: > disable=reimported,unused-import,redefined-outer-name # pylint: > disable=self-assigning-variable # pylint: disable=signature-differs # pylint: > disable=signature-differs # noqa: D403 # pylint: disable=singleton-comparison > # pylint: disable=subprocess-popen-preexec-fn # pylint: > disable=subprocess-popen-preexec-fn,consider-using-with # pylint: > disable=subprocess-run-check # pylint: disable=super-init-not-called # > pylint: disable=syntax-error # pylint: disable=too-few-public-methods # > pylint: disable=too-many-ancestors # pylint: disable=too-many-arguments # > pylint: disable=too-many-arguments,inconsistent-return-statements # pylint: > disable=too-many-arguments, too-many-locals # pylint: > disable=too-many-arguments,too-many-locals # pylint: > disable=too-many-arguments,too-many-locals,too-many-branches # pylint: > disable=too-many-arguments,too-many-locals, too-many-statements # pylint: > disable=too-many-boolean-expressions # pylint: disable=too-many-branches # > pylint: disable=too-many-function-args # pylint: > disable=too-many-instance-attributes # pylint: > disable=too-many-instance-attributes, too-many-arguments # pylint: > disable=too-many-instance-attributes,too-many-arguments,too-many-branches # > pylint: disable=too-many-instance-attributes,too-many-locals # pylint: > disable=too-many-instance-attributes,too-many-public-methods # pylint: > disable=too-many-lines # pylint: disable=too-many-locals # pylint: > disable=too-many-locals,too-many-arguments # pylint: > disable=too-many-locals,too-many-arguments,invalid-name # pylint: > disable=too-many-locals,too-many-arguments, too-many-branches # pylint: > disable=too-many-locals,too-many-statements # pylint: > disable=too-many-nested-blocks # pylint: disable=too-many-public-methods # > pylint: disable=too-many-return-statements # pylint: > disable=too-many-statements # pylint: disable=undefined-variable # pylint: > disable=unexpected-keyword-arg # pylint: disable=ungrouped-imports # pylint: > disable=unidiomatic-typecheck # pylint: disable=unnecessary-lambda # pylint: > disable=unsubscriptable-object # pylint: disable=unsupported-membership-test > # pylint: disable = unused-argument # pylint: disable=unused-argument # > pylint: disable=unused-argument,invalid-name # pylint: > disable=unused-argument,line-too-long # pylint: disable=unused-argument # > pragma: no cover # pylint: > disable=unused-argument,too-many-arguments,too-many-locals # pylint: > disable=unused-argument, unused-variable # pylint: disable=unused-import # > pylint: disable=unused-import,line-too-long # pylint: disable=unused-import # > noqa # pylint: disable=unused-import # noqa: F401 # pylint: > disable=unused-import # noqa: F401 # pylint: disable=unused-variable # > pylint: disable=unused-variable # noqa # pylint: disable=using-constant-test > # pylint: disable=W0104 # pylint: disable=W0106 # pylint: disable=W0108 # > pylint: disable=W0143 # pylint: disable=W0201 # pylint: disable=W0212 # > pylint: disable=W0612 # pylint: disable=W0703 # pylint: > disable=wildcard-import # pylint: disable=wrong-import-order # pylint: > disable=wrong-import-position On Thu, Jun 24, 2021 at 1:57 AM Tomasz Urbaszek > <turbas...@apache.org> wrote: > > -0.5 from me. Black and flake8 does not > guarantee the same quality and consistency of code as pylint (for example > using if list instead of if len(list) > 0 ). > > > I can't remember the last > time where pylint prevented any real error > > Well, isn't it a survivor bias > (the one usually represented by this picture of airplane)? > > Pylint is slow > and sometimes brittle but I think it may help. What I would consider is to > keep pylint for providers - this code in particular is created and changed by > plenty of people with different backgrounds and standards. > > Tomek > > On > Wed, 23 Jun 2021 at 21:45, Jarek Potiuk <ja...@potiuk.com> wrote: >> >> Just > to be precise - the vote will end on Monday 28th at 8.30 PM CEST >> (to > account for weekend) ... >> >> J. >> >> On Wed, Jun 23, 2021 at 9:29 PM Kaxil > Naik <kaxiln...@gmail.com> wrote: >> > >> > +1 -- Pylint is proving to be a > headache and is very very slow locally this days >> > >> > On Wed, Jun 23, > 2021 at 8:02 PM Jarek Potiuk <ja...@potiuk.com> wrote: >> >> >> >> Why not > +10000 Ash :) ? >> >> >> >> On Wed, Jun 23, 2021 at 9:00 PM Andrew Godwin >> > >> <andrew.god...@astronomer.io.invalid> wrote: >> >> > >> >> > I can't cast > a binding vote on this, but I would also like to chime in with support for > this move. We have plenty of protection from everything else, and while we > could try to go through pylint and disable every superfluous check, I'm not > sure it's worth it. >> >> > >> >> > Andrew >> >> > >> >> > On Wed, Jun 23, > 2021 at 12:49 PM Ash Berlin-Taylor <a...@apache.org> wrote: >> >> >> >> >> >> > Clearly I'm +7000 on this. >> >> >> >> >> >> On 23 June 2021 19:38:13 BST, > 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 >> >> >> >> -- >> +48 660 > 796 129 -- +48 660 796 129 -- +48 660 796 129 > > -- > +48 660 796 129 -- +48 660 796 129
