Skip to content

bug_report.md

Latest commit

 

History

History
80 lines (53 loc) · 2.6 KB

bug_report.md

File metadata and controls

80 lines (53 loc) · 2.6 KB
Raw

Bug Report: np.ndarray subscripted type annotation blocks all imports

Severity

CRITICAL — The package is completely unimportable.

Environment

  • Python: 3.9.7
  • NumPy: 1.20.3
  • tdscha: 1.6.2 (main branch)
  • OS: Linux

Description

np.ndarray does not support subscript notation ([...]) — it is not a generic type and lacks __class_getitem__. On line 5032 of Modules/DynamicalLanczos.py, the annotation np.ndarray[np.float64] is used as a parameter type hint:

def get_green_function_continued_fraction(self, w_array : np.ndarray[np.float64], ...):

Because Python evaluates type annotations at definition time (unless from __future__ import annotations is active), this raises:

TypeError: 'type' object is not subscriptable

The error occurs at class definition time, meaning any import of tdscha fails immediately.

Steps to Reproduce

python -c "import tdscha"

Observed output:

TypeError: 'type' object is not subscriptable
  File ".../tdscha/DynamicalLanczos.py", line 5032, in Lanczos
    def get_green_function_continued_fraction(self, w_array : np.ndarray[np.float64], ...):

Root Cause

np.ndarray is a concrete type, not a generic. NumPy provides numpy.typing.NDArray for subscriptable array annotations, but the code uses np.ndarray directly in subscript form, which is syntactically invalid.

Affected Code

File: Modules/DynamicalLanczos.py (installed as tdscha/DynamicalLanczos.py) Line: 5032

def get_green_function_continued_fraction(self, w_array : np.ndarray[np.float64], ...):

Additionally, lines 5388 and 5424 use np.float64 as parameter/return type annotations, which are syntactically valid but unnecessarily restrictive (a plain float would be more appropriate).

Suggested Fixes

Option A (simplest — single line)

Add from __future__ import annotations to the top of Modules/DynamicalLanczos.py. This makes all annotations lazy strings (PEP 563), bypassing the runtime evaluation entirely:

from __future__ import print_function, division, annotations

This is backward-compatible with Python 3.8+ and requires no other code changes.

Option B (best practice)

Replace the invalid subscript annotation with the proper NumPy typing idiom:

from numpy.typing import NDArray

def get_green_function_continued_fraction(self, w_array: NDArray[np.float64], ...):

And optionally change np.float64 annotations on lines 5388 and 5424 to float for broader compatibility.

Workaround

None — the package cannot be imported at all until this is fixed.