Mailman 3 March 2020 - code-quality

Need a Rosetta Stone re PyLint
by Stephen Satchell April 17, 2020

April 17, 2020

I reported a problem with PyLint not knowing that __doc__ is indeed an attribute of a function. So I tried to go through and find out where PyLint makes that determination, between its own code and Astroid code. I'm lost. Is there a roadmap showing where to look for the determination that a particular member is present in a function? I want to develop a patch to fix my problem, I do.

2 1

PyLint's Travis CI-env localy for testing?
by Frank Harrison March 27, 2020

March 27, 2020

Hey all, I have a patch to fix duplicate code detection with multiple jobs in pylint. Right now I have a CI failure (Windows and Ubuntu) that I am unable to repro locally (OSX and Fedora). I would like to mirror the travis linux runner setup on one of my local machines but am unsure of how to do that. On travis it looks like we use straight-up ubuntu image. Is there a nice/fast way to get that and all the pylint environment locally? NB: I have a suspicion that the issue is a combination of process-ordering differences between machines/platforms, the exit code being bitwise or'd with the results from the checkers and the fact that SimilarChecker now works correctly in a multi-proc/jobbed config - causing bit 8 to be or'd with bit 2 in certain conditions. If I can repro locally I will attempt to make the behaviour deterministic. Thanks! -- Frank Harrison double the fish ltd. Kemp House 160 City Road London United Kingdom EC1V 2NX Tel: 07854616533 Email: frank(a)doublethefish.com

1 0

Pylint giving strange error message
by Stephen Satchell March 14, 2020

March 14, 2020

I'm in the process of doing some dynamic documentation with a series of functions. 1. I tried using a .format() method on the doc string, and it just plain didn't work -- to the point that function.__doc__ was not set at all. 2. I tried dynamically appending the dynamic portion of the documentation using function.__doc__ += DYNAMIC_DOC, and it works when run, but Pylint complains E:000, 0: Function 'function' has no '__doc__' member (no-member) Yet, a dir(function) shows that there is indeed a __doc__ attribute. I know this is a little unusual, but I'm trying to maintain a REST application with lots of endpoints and some really silly non-harmonious parameter --and no one available to fix the nightmare in the calling packages! So, it runs. Would a Pylint developer look to see if Pylint is wrong about what a function has in the way of members?

1 0

[bug]broken dependencies with wrapt 1.12.0
by 中川祥 March 6, 2020

March 6, 2020

astroid 2.3.3 on pip needs wrapt 1.11.*, but latest wrapt is 1.12.0. So `pip check` returns an error.

2 1

pep8
by Demetrio Rodriguez Tereshkin March 5, 2020

March 5, 2020

Dear PyCQA, I have a very simple question to which I am unable to find a simple an clear answer: *Does pylint fully cover pep8?* Thank you in advance for clarifying! Best, Demetrio *--* *Demetrio Rodriguez Tereshkin* Machine Learning Engineer +43 664 4846744 <+436644846744> demetrio.rodriguez.tereshkin(a)sclable.com *Sclable Business Solutions GmbH* Marc-Aurel-Straße 10-12/Top 10, 1010 Vienna, Austria https://sclable.com Company Registration Number: FN 387162g Commercial Register Court: Commercial Court of Vienna

3 4

Pip borks the #! line when installing pylint from Python sandbox
by Skip Montanaro March 3, 2020

March 3, 2020

I'm working on Python from GitHub, so have a fork which I check out and keep synced with the main repo. I run pylint on my code (which is nominally aimed at finding its way into Python someday), so installed it using pip: ./python -m pip install --user pylint (I prefer to tie pylint to the Python version I'm working on, if for no other reason than that as Python 3.9alpha moves along, this might flag something needing the Pylint team's attention.) This downloads whatever is necessary and installs it. Unfortunately, instead of creating a #! line of something like #!/home/skip/src/python/cpython/python it generates #!/usr/local/bin/python3.9 I didn't specify any sort of --prefix location when I configured, as I don't intend to install the code I'm working on. Mostly I just run "make test." I saw nothing like an --inplace flag in either Python's configure or pip's help output. It's not at all clear that I should give --prefix=$(pwd) to pip as I think that would install all subsidiary modules (pylint, astroid, etc) inside my Python sandbox. Is there some way coax pip into referencing my sandbox python executable, or will I be forced to edit the relevant command scripts once they are installed in ~/.local/bin? Thanks, Skip Montanaro

1 0