fix: enable flake8-bugbear lints and fix found problems

[cpcloud]

Enable flake8-bugbear ruff lints and fix problems.

[copy-pr-bot]

Auto-sync is disabled for ready for review pull requests in this repository. Workflows must be run manually.

Contributors can view more details about this message here.

[cpcloud]

/ok to test

[cpcloud]

Actually a decent amount of unnecessary API calls happening in these modules, mostly enumerates and pulling out keys or values of dicts without using them.

[greptile-apps]

Greptile Overview

Greptile Summary

enabled flake8-bugbear linting rules and fixed all identified issues across 29 Python files.

The changes primarily address:

• B007: Removed unused loop variables by using _ placeholder, .values() instead of .items(), or adding noqa comments where the variable is used after the loop
• B009/B010: Replaced getattr() and setattr() with direct attribute access where safe
• B004: Replaced hasattr(value, "__call__") with callable(value)

All fixes are semantically equivalent to the original code. The two instances with noqa: B007 comments (interpreter.py and deviceufunc.py) correctly preserve loop variables that are used after the loop completes.

Confidence Score: 5/5

• This PR is safe to merge with minimal risk
• All changes are automated linting fixes that maintain semantic equivalence. The refactorings (removing unused variables, replacing getattr/setattr with direct access, using callable() builtin) are standard Python best practices. The two noqa comments are correctly placed where loop variables are genuinely used after the loop.
• No files require special attention

Important Files Changed

File Analysis

Filename	Score	Overview
pyproject.toml	5/5	enabled flake8-bugbear rules B007 and added ignores for B028, B904, B905; formatting changes to improve readability
numba_cuda/numba/cuda/core/boxing.py	5/5	replaced setattr() calls with direct attribute assignment (B009/B010 fix)
numba_cuda/numba/cuda/core/bytecode.py	5/5	replaced loop with list comprehension to remove unused variable (B007 fix)
numba_cuda/numba/cuda/core/byteflow.py	5/5	removed unused loop variables by using .values() or _ placeholder (B007 fix)
numba_cuda/numba/cuda/core/interpreter.py	5/5	added noqa: B007 comment for loop variable i that is used after the loop
numba_cuda/numba/cuda/core/transforms.py	5/5	replaced manual loop with range object and removed unused loop variable (B007 fix)
numba_cuda/numba/cuda/cudaimpl.py	5/5	optimized strides computation by using in-place reverse instead of double list reversal (B007 fix)
numba_cuda/numba/cuda/deviceufunc.py	5/5	added noqa: B007 for loop variable i used in exception handler; replaced getattr() with direct attribute access (B009 fix)
numba_cuda/numba/cuda/tests/cudapy/test_vectorize.py	5/5	replaced setattr() calls with direct attribute assignment (B010 fix)
numba_cuda/numba/cuda/typing/templates.py	5/5	replaced getattr() calls with direct attribute access (B009 fix)

[greptile-apps]

Greptile Overview

Greptile Summary

This PR enables flake8-bugbear lints in the ruff configuration and systematically fixes all flagged issues across 46 files. The changes are primarily code quality improvements with no behavioral modifications.

Key improvements:

• Fixed mutable default arguments (B006) - Changed [] to () and {} to None in function signatures
• Eliminated loop variable capture bugs (B023) - Functions defined in loops now explicitly receive loop variables as parameters
• Replaced unused loop variables with _ (B007) - Improved code clarity
• Removed unnecessary getattr() and setattr() calls (B009, B010) - Using direct attribute access
• Improved code efficiency - Changed list to set for membership checks, used generator expressions
• Added strategic noqa comments where lint warnings are intentional (e.g., B018 for side-effect attribute access in cloudpickle)
• Removed dead test code - Deleted already-skipped TestFailingStream class

Configuration changes:

• Enabled flake8-bugbear rules ("B") in pyproject.toml
• Added specific rule exceptions (B028, B904, B905, B019) for project-wide patterns
• Added per-file exceptions for test files (B007, B018, B023)
• Updated ruff to >=0.14.10 in pixi.toml

All changes are defensive improvements that eliminate potential bugs without altering functionality.

Confidence Score: 5/5

• This PR is safe to merge with minimal risk - all changes are defensive code quality improvements
• The PR exclusively consists of automated lint fixes addressing flake8-bugbear warnings. Key fixes include eliminating mutable default arguments (common Python gotcha), fixing loop variable capture in closures (prevents subtle bugs), and removing unnecessary code patterns. All changes improve code quality without altering behavior. The systematic nature of the changes and focus on well-known anti-patterns make this a low-risk PR.
• No files require special attention

Important Files Changed

File Analysis

Filename	Score	Overview
pyproject.toml	5/5	Enabled flake8-bugbear lints (B rules) and added specific rule exceptions; reformatted with more readable line breaks
pixi.toml	5/5	Added ruff >=0.14.10 dependency and dev environment for linting work
numba_cuda/numba/cuda/core/interpreter.py	5/5	Fixed loop variable capture issues in closures by passing parameters explicitly; added noqa comments where needed
numba_cuda/numba/cuda/decorators.py	5/5	Changed mutable default argument from [] to () in jit decorator
numba_cuda/numba/cuda/extending.py	5/5	Fixed mutable default argument from {} to None with proper handling in function body
numba_cuda/numba/cuda/typing/templates.py	5/5	Replaced unnecessary getattr() calls with direct attribute access
numba_cuda/numba/cuda/core/ir.py	5/5	Improved loop efficiency, changed list to set for membership checks, fixed closure variable capture
numba_cuda/numba/cuda/core/byteflow.py	5/5	Replaced unused loop variables with _, changed append loop to extend with generator for efficiency
numba_cuda/numba/cuda/deviceufunc.py	5/5	Fixed mutable default arguments, replaced getattr() with direct access, modernized string formatting
numba_cuda/numba/cuda/np/arrayobj.py	5/5	Replaced unused loop variables with _, removed unnecessary setattr() call in favor of direct assignment

[cpcloud]

/ok to test

[greptile-apps]

Greptile Overview

Greptile Summary

This PR systematically enables flake8-bugbear lints and fixes all identified issues across 46 files.

Key changes:

• Enabled flake8-bugbear rules (B-series) in pyproject.toml with specific exceptions for problematic rules (B028, B904, B905, B019) and test files (B007, B018, B023)
• Fixed mutable default arguments (B006) by replacing {} with None and [] with () in function signatures across extending.py, decorators.py, base.py, deviceufunc.py, and test_caching.py
• Replaced unused loop variables with _ (B007) in 30+ files to indicate intentional non-use
• Fixed unnecessary getattr() calls (B009) by using direct attribute access in templates.py
• Fixed setattr() usage (B010) by using direct assignment in arrayobj.py
• Fixed function-in-loop closure issue (B023) in interpreter.py by moving nested function outside loop and passing required parameters
• Added # noqa comments where violations are intentional (e.g., checking cell_contents for side effects)
• Removed permanently skipped test class in test_streams.py
• Updated formatting and added ruff to development environment

All changes follow best practices and improve code quality without altering functionality.

Confidence Score: 5/5

• This PR is safe to merge with no identified issues
• All changes are automated linting fixes that follow established best practices. The modifications are systematic (replacing unused variables, fixing mutable defaults, simplifying attribute access) and do not alter program behavior. The fixes address real code quality issues identified by flake8-bugbear. Appropriate # noqa comments are used where violations are intentional.
• No files require special attention

Important Files Changed

File Analysis

Filename	Score	Overview
pyproject.toml	5/5	enabled flake8-bugbear lints (B rules) and added specific rule exceptions for tests, formatted arrays more compactly
numba_cuda/numba/cuda/extending.py	5/5	fixed mutable default argument in overload function from {} to None (B006)
numba_cuda/numba/cuda/decorators.py	5/5	fixed mutable default argument from [] to () for link parameter (B006)
numba_cuda/numba/cuda/core/base.py	5/5	fixed mutable default arguments in compile functions from {} to None with proper handling (B006)
numba_cuda/numba/cuda/core/interpreter.py	5/5	fixed function-in-loop by passing blk parameter, moved do_change outside loop scope, added noqa for necessary cases (B023, B007)
numba_cuda/numba/cuda/deviceufunc.py	5/5	fixed mutable defaults, replaced .format() with f-string, added noqa for necessary loop variable usage (B006, B007)
numba_cuda/numba/cuda/typing/templates.py	5/5	replaced unnecessary getattr calls with direct attribute access (B009)
numba_cuda/numba/cuda/np/arrayobj.py	5/5	replaced unused loop variables, simplified setattr to direct assignment (B007, B010)
numba_cuda/numba/cuda/tests/cudadrv/test_streams.py	5/5	removed unused test class that was permanently skipped
numba_cuda/numba/cuda/tests/cudapy/test_caching.py	5/5	removed unused envvars parameter and its usage (B006)

[kkraus14]

Should these throw instead of asserting as well?

[cpcloud]

Depends on whether these are validating input for users or making assertions about invariants. If they are asserting invariants then I would leave them. If users can provide "typical" inputs that can violate these assertions then we should make them proper raises.

[cpcloud]

/ok to test

[greptile-apps]

Greptile Summary

This PR systematically enables flake8-bugbear lints and fixes all identified issues across 45 files. The changes focus on improving code quality by addressing common Python pitfalls.

Key Changes:

• Configuration: Enabled flake8-bugbear lints with selective ignores (B028, B904, B905, B019) and per-file ignores for test files (B007, B018, B023)
• Mutable default arguments (B006): Fixed all instances of mutable defaults (list=[], dict={}) by replacing with None or immutable types (tuple=())
• Unnecessary getattr/setattr (B009/B010): Replaced getattr(obj, "attr") with direct attribute access obj.attr and removed unnecessary setattr calls
• Loop variable issues (B007/B023): Fixed unused loop variables with underscore and resolved function-in-loop closures by passing variables as parameters
• Empty pass statements (B027): Removed empty pass from abstract methods that don't need implementation
• Code style improvements: Converted loops to comprehensions/generators where appropriate, changed lists to sets for membership checks

Impact:
All changes are non-functional improvements that maintain existing behavior while following Python best practices. The fixes address potential bugs (mutable defaults, closure issues) and improve code readability and performance (direct attribute access, appropriate data structures).

Confidence Score: 5/5

• This PR is safe to merge with minimal risk
• All changes are automated lint fixes following established Python best practices. The fixes address real code quality issues (mutable defaults, closure bugs) without changing functionality. Changes are well-tested across 45 files with consistent patterns and appropriate use of noqa comments where intentional.
• No files require special attention

Important Files Changed

Filename	Overview
pyproject.toml	Enabled flake8-bugbear lints with specific rules ignored and reformatted for consistency
pixi.toml	Added ruff to dev environment and reformatted task configuration
numba_cuda/numba/cuda/decorators.py	Changed mutable default argument from list to tuple in jit decorator (B006)
numba_cuda/numba/cuda/extending.py	Fixed mutable default dict argument in overload function (B006)
numba_cuda/numba/cuda/deviceufunc.py	Fixed mutable defaults, replaced getattr with direct access, improved f-string formatting
numba_cuda/numba/cuda/core/interpreter.py	Fixed loop variable closure issue by passing block as parameter (B023), added noqa for intentional unused loop var
numba_cuda/numba/cuda/core/bytecode.py	Replaced loop with list comprehension to avoid unused loop variable (B007)
numba_cuda/numba/cuda/core/ir.py	Fixed loop variable closure (B023), optimized count_spaces with generator, changed list to set for membership checks
numba_cuda/numba/cuda/typing/templates.py	Replaced unnecessary getattr with direct attribute access (B009)
numba_cuda/numba/cuda/np/arrayobj.py	Replaced unused loop variables with underscore, removed unnecessary setattr (B010)