Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Make FunctionDef.implicit_parameters return 1 for methods #1531

Merged

Conversation

jacobtylerwalls
Copy link
Member

@jacobtylerwalls jacobtylerwalls commented Apr 26, 2022

Description

When a node is inferred as FunctionDef instead of BoundMethod, implicit_parameters() always returns 0, which can cause problems in pylint with no-value-for-parameter and such. We can at least fall back to the knowledge that it is bound and return 1.

Type of Changes

Type
🐛 Bug fix

Related Issue

Closes pylint-dev/pylint#6464 (because we will not be writing Qt tests there)

  • Todo: see if this closes any other pylint issues for no-value-for-parameter

Relatedly, make `FunctionDef.is_bound` return True for instance methods.
@jacobtylerwalls
Copy link
Member Author

@The-Compiler does this resolve the false positive for you?

@jacobtylerwalls jacobtylerwalls changed the title Make FunctionDef.implicit_parameters return 1 for instance methods Make FunctionDef.implicit_parameters return 1 for methods Apr 26, 2022
@@ -1642,7 +1645,7 @@ def is_bound(self):
False otherwise.
:rtype: bool
"""
return self.type == "classmethod"
return self.type in {"method", "classmethod"}
Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

This method was untested according to coveralls.

@coveralls
Copy link

coveralls commented Apr 26, 2022

Pull Request Test Coverage Report for Build 2250456664

  • 3 of 3 (100.0%) changed or added relevant lines in 1 file are covered.
  • No unchanged relevant lines lost coverage.
  • Overall coverage increased (+0.02%) to 91.634%

Totals Coverage Status
Change from base Build 2250330420: 0.02%
Covered Lines: 9135
Relevant Lines: 9969

💛 - Coveralls

Copy link
Member

@Pierre-Sassoulas Pierre-Sassoulas left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

LGTM !

@jacobtylerwalls
Copy link
Member Author

Todo: see if this closes any other pylint issues for no-value-for-parameter

I can't say I checked every one, but I checked multiple and this didn't seem to fix any.

@jacobtylerwalls jacobtylerwalls added the pylint-tested PRs that don't cause major regressions with pylint label Apr 27, 2022
Comment on lines +51 to +52
if attribute_node is Uninferable:
pytest.skip("PyQt6 C bindings may not be installed?")
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Don't we want to force the c binding to work correctly in at least one of the CI pipeline ? Otherwise this test can work fine online and fail for someone with QT locally. I think we could create a pytest mark for "qt" and a CI job with a proper installation.

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Yes, it works in the "release tests" workflow.

Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Hmm, it means we're going to catch issues at release time right ? When/how are we launching this workflow ?

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Just manually with a reminder in the release.md.

Installing qt takes 39s. Unless there's a way to cache it?

Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

For 40s we might as well add it for each commit (one interpreter only ?). There's a lot of available runner for astroid and they run in a reasonable time (compared to pylint anyway).

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Actually, it takes 8s, duh, because most of that linked run was setting up the env!

Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I'm confused - why does "Install Qt" actually only do sudo apt-get install build-essential libgl1-mesa-dev? You are installing PyQt from binary wheels, so that 40s step can probably just be dropped entirely.

Copy link
Member Author

@jacobtylerwalls jacobtylerwalls Apr 28, 2022

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

We discovered that the PyQt6 wheel doesn't necessarily install the C bindings on ubuntu-latest, see failing run and comment.

@Pierre-Sassoulas Pierre-Sassoulas dismissed their stale review April 27, 2022 17:52

Realized that one job was skipped if QT6 was not properly installed.

astroid/nodes/scoped_nodes/scoped_nodes.py Outdated Show resolved Hide resolved
tests/unittest_brain_qt.py Outdated Show resolved Hide resolved
@The-Compiler
Copy link
Contributor

The-Compiler commented Apr 28, 2022

@The-Compiler does this resolve the false positive for you?

Almost! It still complains on two lines compared to 2.13.7:

https://github.com/qutebrowser/qutebrowser/blob/6fc30a1e851dbc07461ba50d4266bbcae40e1fb1/qutebrowser/misc/throttle.py#L88

and

https://github.com/qutebrowser/qutebrowser/blob/6fc30a1e851dbc07461ba50d4266bbcae40e1fb1/tests/unit/config/test_configtypes.py#L226

The former sounds like pylint/astroid thinks a slot is needed for .disconnect(), but that's not the case, it can be called without one. I think that's a bug in the Qt brain here:

https://github.com/PyCQA/astroid/blob/8fda6c667a11050d36b267ded157933bbafb8290/astroid/brain/brain_qt.py#L30

(should be slot=None, same for PySide most likely).

The latter gives me too-many-function-args and is probably an unrelated false-positive regression.

@jacobtylerwalls
Copy link
Member Author

Thanks for testing. Feel free to open new issues or PRs. Apparently slot is optional, but you cannot pass an explicit None, see python-qt-tools/PyQt5-stubs#103. I'm a little hesitant to show that slot=None is the default value in that case. See also pylint-dev/pylint#5264 for general weakness in this area.

@jacobtylerwalls jacobtylerwalls merged commit 9968101 into pylint-dev:main Apr 30, 2022
@jacobtylerwalls jacobtylerwalls deleted the qt-no-value-for-param branch April 30, 2022 15:57
The-Compiler added a commit to The-Compiler/astroid that referenced this pull request May 10, 2022
Discussed here:
pylint-dev#1531 (comment)

PyQt supports calling .disconnect() without any arguments in order to
disconnect all slots:
https://www.riverbankcomputing.com/static/Docs/PyQt6/signals_slots.html#disconnect

Strictly speaking, slot=None is a wrong call, as actually passing None
does not work:
python-qt-tools/PyQt5-stubs#103

However, pylint/astroid does not support overloads needed to properly
express this: pylint-dev/pylint#5264

So, while this is "wrong", it's less wrong than before - without this
change, pylint expects a mandatory argument, thus raising a
false-positive here:

    from PyQt5.QtCore import QTimer
    t = QTimer()
    t.timeout.connect(lambda: None)
    t.timeout.disconnect()

despite running fine, pylint complains:

    test.py:4:0: E1120: No value for argument 'slot' in method call (no-value-for-parameter)

while with this change, things work fine.
Pierre-Sassoulas pushed a commit that referenced this pull request May 12, 2022
* qt brain: Make slot argument optional for disconnect()

Discussed here:
#1531 (comment)

PyQt supports calling .disconnect() without any arguments in order to
disconnect all slots:
https://www.riverbankcomputing.com/static/Docs/PyQt6/signals_slots.html#disconnect

Strictly speaking, slot=None is a wrong call, as actually passing None
does not work:
python-qt-tools/PyQt5-stubs#103

However, pylint/astroid does not support overloads needed to properly
express this: pylint-dev/pylint#5264

So, while this is "wrong", it's less wrong than before - without this
change, pylint expects a mandatory argument, thus raising a
false-positive here:

    from PyQt5.QtCore import QTimer
    t = QTimer()
    t.timeout.connect(lambda: None)
    t.timeout.disconnect()

despite running fine, pylint complains:

    test.py:4:0: E1120: No value for argument 'slot' in method call (no-value-for-parameter)

while with this change, things work fine.
Pierre-Sassoulas pushed a commit that referenced this pull request Jun 13, 2022
* qt brain: Make slot argument optional for disconnect()

Discussed here:
#1531 (comment)

PyQt supports calling .disconnect() without any arguments in order to
disconnect all slots:
https://www.riverbankcomputing.com/static/Docs/PyQt6/signals_slots.html#disconnect

Strictly speaking, slot=None is a wrong call, as actually passing None
does not work:
python-qt-tools/PyQt5-stubs#103

However, pylint/astroid does not support overloads needed to properly
express this: pylint-dev/pylint#5264

So, while this is "wrong", it's less wrong than before - without this
change, pylint expects a mandatory argument, thus raising a
false-positive here:

    from PyQt5.QtCore import QTimer
    t = QTimer()
    t.timeout.connect(lambda: None)
    t.timeout.disconnect()

despite running fine, pylint complains:

    test.py:4:0: E1120: No value for argument 'slot' in method call (no-value-for-parameter)

while with this change, things work fine.
Pierre-Sassoulas pushed a commit that referenced this pull request Jun 13, 2022
* qt brain: Make slot argument optional for disconnect()

Discussed here:
#1531 (comment)

PyQt supports calling .disconnect() without any arguments in order to
disconnect all slots:
https://www.riverbankcomputing.com/static/Docs/PyQt6/signals_slots.html#disconnect

Strictly speaking, slot=None is a wrong call, as actually passing None
does not work:
python-qt-tools/PyQt5-stubs#103

However, pylint/astroid does not support overloads needed to properly
express this: pylint-dev/pylint#5264

So, while this is "wrong", it's less wrong than before - without this
change, pylint expects a mandatory argument, thus raising a
false-positive here:

    from PyQt5.QtCore import QTimer
    t = QTimer()
    t.timeout.connect(lambda: None)
    t.timeout.disconnect()

despite running fine, pylint complains:

    test.py:4:0: E1120: No value for argument 'slot' in method call (no-value-for-parameter)

while with this change, things work fine.
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
Bug 🪳 pylint-tested PRs that don't cause major regressions with pylint
Projects
None yet
Development

Successfully merging this pull request may close these issues.

False positive no-value-for-parameter for Qt methods
5 participants