Skip to content

docs-knownissues-python.md

Latest commit

 

History

History
292 lines (219 loc) · 10.8 KB

docs-knownissues-python.md

File metadata and controls

292 lines (219 loc) · 10.8 KB
Raw

knownissues-python

Contact: numbworks@gmail.com

Revision History

Date Author Description
2026-05-18 numbworks Created.
2026-06-28 numbworks Last update.

Introduction

This documents collects all the known issues related to some or all the nw applications written in Python.

Known Issues: "ImportError: cannot import name 'display' from 'IPython.core.display'"

The devcontainer's Dockerfile forces ipkernel to use a specific version of its ipython dependency:

FROM python:3.12.5-bookworm

# ...
RUN pip install ipykernel==6.29.5 ipython==7.23.1
# ...

Without this enforcement in place, the devcontainer would return the following error message:

ImportError: cannot import name 'display' from 'IPython.core.display'

By investigating the dependency tree with the following commands:

pip install pipdeptree
pipdeptree -p ipykernel

I discovered that ipykernel has the following (very lousy) dependency constraint:

...
├── ipython [required: >=7.23.1, ...
...

The >= means that, even if your devcontainer uses a frozen version of ipykernel, the latest version of ipython will be downloaded at each rebuild.

When I started getting the error message, the installed ipython version was 9.2.0 (quite far from the original 7.23.1).

Forcing pip to use an older dependency was necessary to bring back the devcontainer to a working status.

Known Issues: "ModuleNotFoundError: No module named 'tinycss2.color5'"

At the moment of writing, the latest version of weasyprint is 67.0, but it returns the following error if run via devcontainer:

2025-12-22 19:40:18.005 [error] Unittest test discovery error for workspace:  /workspaces/nwreadinglist 
 Failed to import test module: nwreadinglisttests
Traceback (most recent call last):
  File "/usr/local/lib/python3.12/unittest/loader.py", line 396, in _find_test_path
    module = self._get_module_from_name(name)
             ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/usr/local/lib/python3.12/unittest/loader.py", line 339, in _get_module_from_name
    __import__(name)
  File "/workspaces/nwreadinglist/tests/nwreadinglisttests.py", line 17, in <module>
    from nwreadinglist import RLCN, DEFINITIONSTR, OPTION, REPORTSTR, _MessageCollection, RLSummary, DefaultPathProvider
  File "/workspaces/nwreadinglist/src/nwreadinglist.py", line 25, in <module>
    from weasyprint import CSS, HTML
  File "/usr/local/lib/python3.12/site-packages/weasyprint/__init__.py", line 440, in <module>
    from .css import preprocess_stylesheet  # noqa: I001, E402
    ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/usr/local/lib/python3.12/site-packages/weasyprint/css/__init__.py", line 32, in <module>
    from . import counters, media_queries
  File "/usr/local/lib/python3.12/site-packages/weasyprint/css/counters.py", line 10, in <module>
    from .tokens import remove_whitespace
  File "/usr/local/lib/python3.12/site-packages/weasyprint/css/tokens.py", line 8, in <module>
    from tinycss2.color5 import parse_color
ModuleNotFoundError: No module named 'tinycss2.color5'

2025-12-22 19:40:18.229 [info] Unittest discovery completed for workspace /workspaces/nwreadinglist

The root cause of the issue is that weasyprint 67.0 forces the installation of tinycss2 1.4.0 (which doesn't support tinycss2.color5), and even asking the Dockerfile to enforce the installation of tinycss2 1.5.0 (which supports tinycss2.color5) doesn't work (tinycss2 stays on 1.4.0).

To solve the issue we use the older weasyprint 66.0:

...
RUN pip install weasyprint==66.0
...

Known Issues: Pandas 2.2.x and its warnings

Pandas 2.2.0 introduced several warnings related to the "Copy-on-Write" feature planned for Pandas 3.0.0, such as:

...
/tmp/onefile_3061176_1779002299_484551/nwreadinglist.py:853: ChainedAssignmentError: A value is trying to be set on a copy of a DataFrame or Series through chained assignment.
When using the Copy-on-Write mode, such chained assignment never works to update the original DataFrame or Series, because the intermediate object on which we are setting values always behaves as a copy.

Try using '.loc[row_indexer, col_indexer] = value' instead, to perform the assignment in a single step.

See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy
...

The goal was to prepare users for the breaking change by flooding the console with warning messages.

This (quite questionable) strategy had a strong impact on my building pipeline based upon Nuitka 2.8.9, because these warnings, which weren't shown at all during development in Visual Studio Code, appeared auto-magically in the terminal while running the compiled application.

The behaviour is confirmed by the following online resources:

My first attempt to solve the problem was to make the code highligted by the warnings compliant with "Copy-on-Write" - e.g.:

  • Example 1

    rl_df = rl_df[column_names]
...

    rl_df = rl_df[column_names].copy()
...

  • Example 2

    rls_by_year_df.drop(labels = RLCN.MONTH, inplace = True, axis = 1)
rls_by_year_df.drop(labels = RLCN.TRENDSYMBOL, inplace = True, axis = 1)
...

    rls_by_year_df = rls_by_year_df.drop(labels = RLCN.MONTH, axis = 1)
rls_by_year_df = rls_by_year_df.drop(labels = RLCN.TRENDSYMBOL, axis = 1)
...

  • Example 3

    filtered_df = filtered_df.loc[condition]
...

    filtered_df = filtered_df.loc[condition].copy()
...

The first attempt didn't work, the warnings kept appearing.

My second attempt to solve the problem was to add the following block of code at the top of the module, but it didn't solve the problem either:

import warnings
import pandas as pd
pd.options.mode.copy_on_write = True

if "__compiled__" in globals():
    warnings.filterwarnings(
      action = "ignore", 
      message = ".*ChainedAssignmentError.*"
    )
else:
    warnings.filterwarnings(
      action = "error", 
      message = ".*ChainedAssignmentError.*"
    )

My third attempt was to downgrade Pandas to the latest available version before 2.2.0, and adjust numpy and pandas-stubs accordingly. This led me to keep these as-is:

Python 3.12.5
nuitka==2.8.9

but to downgrade these:

numpy==2.1.2
pandas==2.2.3
pandas-stubs==2.2.3.241009

to these:

numpy==1.26.4
pandas==2.1.4
pandas-stubs==2.1.4.231227

My third attempt was successful.

Known Issue: "ModuleNotFoundError: No module named 'click'"

The unit test runner in VSCode doesn't work if click, a dependency of spacy, is not explicitly installed:

2026-06-21 12:36:03.116 [info] Started unittest discovery subprocess (execution factory) for workspace /workspaces/nwrefurbishedanalytics
2026-06-21 12:36:03.916 [error] Unittest test discovery error for workspace:  /workspaces/nwrefurbishedanalytics

Failed to import test module: nwrefurbishedanalyticsclitests
Traceback (most recent call last):

  File "/usr/local/lib/python3.12/unittest/loader.py", line 396, in _find_test_path
    module = self._get_module_from_name(name)
             ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/usr/local/lib/python3.12/unittest/loader.py", line 339, in _get_module_from_name
    __import__(name)
  File "/workspaces/nwrefurbishedanalytics/tests/nwrefurbishedanalyticsclitests.py", line 13, in <module>
    from nwrefurbishedanalytics import DEBUGGABLE, FILTER, LOGGABLE, SAVEABLE, SETTINGBAGDEFAULT, ComponentBag
  File "/workspaces/nwrefurbishedanalytics/tests/../src/nwrefurbishedanalytics.py", line 16, in <module>
    import spacy
  File "/usr/local/lib/python3.12/site-packages/spacy/__init__.py", line 16, in <module>
    from .cli.info import info  # noqa: F401
    ^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/usr/local/lib/python3.12/site-packages/spacy/cli/__init__.py", line 4, in <module>
    from . import download as download_module  # noqa: F401
    ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/usr/local/lib/python3.12/site-packages/spacy/cli/download.py", line 19, in <module>
    from ._util import SDIST_SUFFIX, WHEEL_SUFFIX, Arg, Opt, app
  File "/usr/local/lib/python3.12/site-packages/spacy/cli/_util.py", line 22, in <module>
    from click import NoSuchOption

ModuleNotFoundError: No module named 'click'

To solve the issue, explicitly install click before spacy in the devcontainer:

...
RUN pip install click==8.4.1
RUN pip install spacy==3.8.7
...

Known Issue: "RuntimeWarning: Missing required CPU features"

By default, a Python application that has polars as dependency, when compiled and run on an old CPU, throws the following error:

ruben@R-SURFACEGO:~/Downloads/Packages/Internal Releases/NW.ZooSketchDownloader/v2.0.0$ nwrea

/usr/lib/nwrea/polars/_cpu_check.py:255: RuntimeWarning: Missing required CPU features.

The following required CPU features were not detected:
    avx, avx2, fma, bmi1, bmi2
Continuing to use this version of Polars on this processor will likely result in a crash.
Install `polars[rtcompat]` instead of `polars` to run Polars with better compatibility.

Hint: If you are on an Apple ARM machine (e.g. M1) this is likely due to running Python under Rosetta.
It is recommended to install a native version of Python that does not run under Rosetta x86-64 emulation.

If you believe this warning to be a false positive, you can set the `POLARS_SKIP_CPU_CHECK` environment variable to bypass this check.

Illegal instruction

To solve the issue:

  1. install polars in "runtime compatible" mode:

    ...
pip install polars[rtcompat]==1.37.1
...

  2. set the environment variable POLARS_SKIP_CPU_CHECK=1 to prevent Nuitka to trigger its CPU check script during the build phase on your build agent (which most likely has a CPU with AVX) and set two custom compiler flags that will force nuitka to target the baseline 64-bit instruction set from 2004 (which has no AVX requirements):

    ...
export POLARS_SKIP_CPU_CHECK=1 && \
export CCFLAGS="-march=x86-64" && \
export LDFLAGS="-march=x86-64" && \
python3 -m nuitka \
...

  3. Done!

Markdown Toolset

Suggested toolset to view and edit this Markdown file: