pytest assertion rewriting does not seem to work with meson editable install

Aberta
  • Algo errado?
#646 4 comentários 1 reação 0 responsáveis Ver no GitHub

Mantenedores costumam responder em até 1 dia

Ninguém assumiu esta issue ainda.

Avaliação

Dificuldade
4/5
Tempo estimado
3-5 dias
Facilidade para iniciantes
35/100
Tipo de issue
Bug
Clareza
Razoavelmente clara
Status de atividade
Estagnada
Stack de tecnologia
python
Domínio
build-system, testing-qa

Direção de pesquisa

Comece com o reprodutor em numpy/tests/test_mine.py e compare a saída do pytest após os comandos documentados de instalação normal e editable. Rastreie como a instalação editable afeta a reescrita de asserções do pytest, usando como contexto a issue vinculada do Meson. A tarefa estará concluída quando a execução com a instalação editable relatar os valores avaliados das asserções, correspondendo ao comportamento da instalação normal.

Escrita pelo modelo de indexação a partir do texto da issue.

Descrição

dependency-bug

It could well be partially linked to some of the discussion in https://github.com/mesonbuild/meson-python/issues/642 but I thought a separate issue may be worth it.

Description of the issue

The pytest assertion rewriting that shows additional debug information when an assert fails does not show up when using meson editable install. It does work fine when using normal (non-editable) meson install. Although not crucial I would argue that this can be actually quite useful in CI logs.

It could also well be a Pytest issue but I thought I had to start somewhere and chose meson-python.

More background about Pytest assertion rewriting, see Pytest doc that links to a somewhat old blog post but I am guessing is probably still relevant?

This was originally seen in scikit-learn (see https://github.com/scikit-learn/scikit-learn/issues/29253#issuecomment-2173549242 if you are really curious where it would have actually been quite useful to see intermediary values in the AssertionError in https://github.com/scikit-learn/scikit-learn/issues/29253#issuecomment-2167610891). I put together a reproducer below with numpy hoping this makes it a bit easier to look at. I guess a smaller reproducer could be put together if deemed useful, but it would take a bit of time (I tried with https://github.com/dnicolodi/python-siphash24 and I am not able to reproduce the issue).

Get the reproducer from my numpy branch
git clone https://github.com/lesteve/numpy
cd numpy
git checkout meson-editable-pytest-assertion-rewrite

I added a simple test in numpy/tests/test_mine.py

import numpy as np


def test():
    arr = np.arange(100)
    assert arr.sum() < 100

When running with pytest you pytest assertion rewriting shows the value or arr.sum() (see below).

Normal (non-editable) install
pip install --verbose --no-build-isolation .

Note the assertion rewriting i.e. you get the value of arr.sum()

Partial output:

E       assert np.int64(4950) < 100
E        +  where np.int64(4950) = <built-in method sum of numpy.ndarray object at 0x7b77efd7a370>()

Full output

 ❯ pytest --pyargs numpy.tests.test_mine
========================== test session starts ===========================
platform linux -- Python 3.11.7, pytest-8.2.2, pluggy-1.5.0
rootdir: /home/lesteve
plugins: hypothesis-6.103.2
collected 1 item                                                         

micromamba/envs/numpy-dev/lib/python3.11/site-packages/numpy/tests/test_mine.py F [100%]

================================ FAILURES ================================
__________________________________ test __________________________________

    def test():
        arr = np.arange(100)
>       assert arr.sum() < 100
E       assert np.int64(4950) < 100
E        +  where np.int64(4950) = <built-in method sum of numpy.ndarray object at 0x7b77efd7a370>()
E        +    where <built-in method sum of numpy.ndarray object at 0x7b77efd7a370> = array([ 0,  1,  2,  3,  4,  5,  6,  7,  8,  9, 10, 11, 12, 13, 14, 15, 16,\n       17, 18, 19, 20, 21, 22, 23, 24, 25, ...72, 73, 74, 75, 76, 77, 78, 79, 80, 81, 82, 83, 84,\n       85, 86, 87, 88, 89, 90, 91, 92, 93, 94, 95, 96, 97, 98, 99]).sum

micromamba/envs/numpy-dev/lib/python3.11/site-packages/numpy/tests/test_mine.py:6: AssertionError
======================== short test summary info =========================
FAILED micromamba/envs/numpy-dev/lib/python3.11/site-packages/numpy/tests/test_mine.py::test - assert np.int64(4950) < 100
=========================== 1 failed in 0.82s ============================
Editable install
pip install --verbose --no-build-isolation --editable . --config-settings editable-verbose=true

You don't get the value of arr.sum()

 pytest --pyargs numpy.tests.test_mine
========================== test session starts ===========================
platform linux -- Python 3.11.7, pytest-8.2.2, pluggy-1.5.0
rootdir: /home/lesteve
plugins: hypothesis-6.103.2
collecting ... + /home/lesteve/micromamba/envs/numpy-dev/bin/ninja
[1/1] Generating numpy/generate-version with a custom command
Saving version to numpy/version.py
collected 1 item                                                         

dev/numpy/numpy/tests/test_mine.py F                               [100%]

================================ FAILURES ================================
__________________________________ test __________________________________

    def test():
        arr = np.arange(100)
>       assert arr.sum() < 100
E       AssertionError

dev/numpy/numpy/tests/test_mine.py:6: AssertionError
======================== short test summary info =========================
FAILED dev/numpy/numpy/tests/test_mine.py::test - AssertionError
=========================== 1 failed in 1.02s ============================
Linguagem predominante
Python
Estrelas
182
Forks
97
Merge médio
21h 24min
PRs com merge (30d)
8

Preparar o ambiente

Este projeto não oferece contêiner de desenvolvimento, Dockerfile nem guia de contribuição, então a configuração fica por sua conta: comece pelo README e veja nosso guia da primeira contribuição para os passos gerais.

Primeiros passos

  1. Leia a issue inteira e depois o guia de contribuição do projeto.
  2. Comente na issue dizendo que vai assumir — evita que duas pessoas façam o mesmo trabalho.
  3. Faça um fork do repositório e trabalhe em uma branch.
  4. Abra um pull request que referencie o número da issue.

Mais de mesonbuild/meson-python

Todas as issues de mesonbuild/meson-python

Issues semelhantes

Mais issues de Python

Receba novas issues na sua caixa de entrada

Um resumo curto de issues do GitHub para quem está começando.