Check sets for duplicate entries

[nickdrozd]

Current problem

Pylint check dictionaries for duplicates, but not sets.

DICTIONARY = {
    1: 2,
    2: 3,
    1: 2,  # duplicate flagged
}

SET = {
    1,
    2,
    1,  # duplicate not flagged
}

Desired solution

Flag duplicate set entries just as with dictionaries.

Additional context

No response

[ksaketou]

Hello, I'd like to work on that. If I understood correctly, the message we are talking about here is duplicate-key?

[DanielNoord]

Yes! We might also need to update the description of the message to not only mention dictionaries any more 😉

[ksaketou]

Okayy, so we don't have to create a new message for the sets?

[DanielNoord]

Hm, now that I think about it, perhaps we should create duplicate-elements? The original pep refers to the entries in a set as values and elements. Using keys might be a bit confusing? What do others think?

[Pierre-Sassoulas]

I think creating a new message duplicate-values make sense.

[nickdrozd]

The PEP says:

Programmers are often told that they can implement sets as dictionaries with “don’t care” values. Items can be added to these “sets” by assigning the “don’t care” value to them; membership can be tested using dict.has_key...

From this perspective, set entries are more like "keys". But then the rest of the PEP refers to them as "values". And the docstring for set mentions "elements", as is usually done in math.

So in the interest of covering every perspective, I suggest we use duplicate-kvalemuys.

[ksaketou]

Hello, I am trying to run the tests but I get a ModuleNotFoundError at the primer tests for this line:
https://github.com/PyCQA/pylint/blob/8e9a44727545b09ccc0161d72b8b66bc92d95d88/pylint/testutils/primer.py#L5
Any idea why this is happening?

[Pierre-Sassoulas]

Hello @ksaketou, you need to install pylint locally with pip3 install -e .[testutil] we made this package an extra in #5486 in order to not add a runtime dependency for standard users.

[ksaketou]

Hello again, I'm getting an unexpected syntax-error at the symlink_module1 functional test, but I don't know how my changes could be related to this specific error message. I believe this is the line with the problem:
https://github.com/PyCQA/pylint/blob/975550582494f9e1c0eb43336943c716120bc4e6/tests/functional/s/symlink/_binding/symlink_module.py#L1

[DanielNoord]

How did you clone the repository? That functional test is dependent on symlinks so I think it is important that the repository is cloned via git instead of installed via pip or downloaded directly from Github.

[ksaketou]

I cloned it with git clone. The path of this line is correct though, it's not that I miss any files.

[DanielNoord]

Did you get a output log for the syntax error_

[ksaketou]

That's the output I get.

Pylance also points the line as an error with the message Expected expression.

[DanielNoord]

Hm, you might need to reclone your repository. That file should not show ../symlink_module/symlink_module.py but rather the content of ../symlink_module/symlink_module.py , which is:

"""Example taken from issue #1470"""


def func():
    """Both module should be parsed without problem"""
    return 1

I believe git should automatically copy any symlinks, but apparently it failed to do so for you locally. (CI passes on Windows so it should work I guess?...)

The files should show up like this in the VSC File Explorer, to indicate that they are symlinked.

Edit: Some googling showed that this might be relevant:
https://gh.neting.ccmunity/t/git-bash-symbolic-links-on-windows/522/4
https://stackoverflow.com/questions/5917249/git-symlinks-in-windows
https://github-wiki-see.page/m/git-for-windows/git/wiki/Symbolic-Links
https://www.quora.com/How-does-Git-handle-symbolic-links-on-Windows

[ksaketou]

I have two files. The _binding/symlink_module.py contains the path I showed you above and the symlink_module/symlink_module.py contains this:

"""Example taken from issue #1470"""


def func():
    """Both module should be parsed without problem"""
    return 1

And these are the files at my Explorer:

Thank you for the links, I will try to fix this :)

[ksaketou]

Keeping the symlink as a string like below seems to work for my case so far. However, I don't know if there is any problem with changing this file.

[DanielNoord]

Yeah, if the file gets changed then the test won't work anymore. Perhaps you can revert the change and see if you can commit without any changes to the file?

Were you unable to allow git to create symlinks?

[ksaketou]

When I run the symlink_module1 test with that change it passes though.

I changed the core.symlinks of the .gitconfig file to True but that didn't seem to work. If I don't find anything else I will try to commit without that change as you said.

[DanielNoord]

I'll need to see the actual commit to see if nothing actually changes. This is difficult to interpret without seeing the actual changes git picks up on. Luckily we have the CI which should also pick up on this 😄

I'd say go ahead with working on this, we should be able to resolve any issues with this particular test if they arise!