Python lint: Ruff rules for pylint and code complexity #525
Conversation
cclauss
force-pushed
the
ruff-rules-for-pylint
branch
from
8690e74 to
1d06973
Compare
There was a problem hiding this comment.
Thank you for contributing to Python SDK!
- Please can you provide rationale for the specific numerical thresholds chosen
- Is implementing both C90 and PL linting rules simultaneously necessary?
- Please remove the widespread import style changes which aren't central to the linting goals
The subprocess check=False change is great, but should be submitted as a separate PR focused on that specific fix.
A subset of: * modelcontextprotocol#525 % `ruff check --select=PLW1510` ``` 1 PLW1510 [*] subprocess-run-without-check ``` https://docs.astral.sh/ruff/rules/subprocess-run-without-check
cclauss
force-pushed
the
ruff-rules-for-pylint
branch
from
1d06973 to
7628dcb
Compare
|
Please read the git commit when reviewing pull requests.
The current code complexity determines the values set in this PR, with comments showing the recommended values. To find complexity in the codebase, just lower any of these values by one and run - mccabe.max-complexity = 13 # Default is 10
- max-branches = 14 # Default is 12
- max-returns = 7 # Default is 6
- max-statements = 69 # Default is 50
+ mccabe.max-complexity = 24 # Default is 10
+ max-branches = 23 # Default is 12
+ max-returns = 13 # Default is 6
- max-statements = 99 # Default is 50
+ max-statements = 102 # Default is 50I have the RUFF settings to ignore the pylint import rules PLC0414 and PLR0402. |
cclauss
force-pushed
the
ruff-rules-for-pylint
branch
from
191f41b to
1596aca
Compare
This replaces absolute URLs with paths, links the "Roots" column header, and moves anchor definitions for the table header links to the top for organizational purposes. Closes modelcontextprotocol#525 Co-authored-by: Andrew Qu <[email protected]>
felixweinberger
added
the
needs more work
cclauss
force-pushed
the
ruff-rules-for-pylint
branch
from
2d617e2 to
f3540b7
Compare
There was a problem hiding this comment.
While I'm not a fan of ignoring ruff rules, in this case I believe it is justified given we're actually sharpening our lint rules overall and simply making explicit what we were previously implicitly ignoring.
Therefore I think we should land this in the interest of managing growing complexity on the Python SDK.
I've also created an issue for our v2 release where ideally we'd get rid of all ignores here.
| "PL", # Pylint | ||
| "UP", # pyupgrade | ||
| ] | ||
| ignore = ["PERF203", "PLC0415", "PLR0402"] |
There was a problem hiding this comment.
| [tool.ruff.lint.per-file-ignores] | ||
| "__init__.py" = ["F401"] | ||
| "tests/server/fastmcp/test_func_metadata.py" = ["E501"] | ||
| "tests/shared/test_progress_notifications.py" = ["PLW0603"] |
There was a problem hiding this comment.
Not my favorite solution, but assuming the same thing applies as above - we have this code already, we were just ignoring it by omission.
felixweinberger
dismissed
ihrpr’s stale review
Requested changes have been addressed
|
@cclauss thank you for the contribution and apologies again for the delay! |
felixweinberger
merged commit c47c767
into
modelcontextprotocol:main
Add ruff rules for Pylint and McCabe cyclomatic code complexity.
%
ruff check --select=C90,PL --statistics | sort -k2Motivation and Context
Pylint checks for errors, enforces a coding standard, looks for code smells, and can suggest how code could be refactored.
McCabe detects functions with high complexity that are hard to understand and maintain.
The C90 and PLR091 rules enable the setting of upper limits on code complexity. If a contributor wants to raise these limits, that can lead to useful discussion in code reviews about why complexity needs to rise and maintainability needs to go down.
How Has This Been Tested?
Breaking Changes
Types of changes
Checklist
Additional context