rapidsai/cudf

[BUG] Tests of ufuncs should ensure values are in the right domain

Open

#12352 opened on Dec 9, 2022

View on GitHub
 (1 comment) (0 reactions) (0 assignees)C++ (6,000 stars) (735 forks)batch import
0 - BacklogPythonbuggood first issuetests

Description

Describe the bug The test_array_ufunc.py module contains tests that I introduced in #10346 and #10217 that are intended to be generic for all ufuncs in test_ufunc_(index|series|dataframe). However, the inputs are generated in exactly the same way for all ufuncs. This leads to some tests where all inputs are invalid, making some of the tests do nothing other than ensure that invalid inputs produce the same output. Some of the most egregious examples are inverse trigonometric functions arcsin and arccos, for which the domains are limited to (-1, 1) but all inputs are > 1.

Steps/Code to reproduce bug Run the tests and check warnings.

Expected behavior The tests should be rewritten to generate inputs within the domain of the functions of choice, with perhaps just a few values outside the domain so that we can continue to validate consistent behavior in both cases. In an ideal world we would also update cudf to throw the same warnings as pandas here, but that may be more work than it's worth since we would almost certainly have to introspect the data.

Contributor guide

Tech stack
pythoncpppandasnumpy
Domain
testing
Issue type
bug
DifficultyEstimated implementation difficulty for a new contributor, from 1 for very small changes to 5 for expert-level work.
3
Estimated timeA rough time range for an experienced contributor to investigate, implement, test, and prepare a pull request.
1-3 hours
Activity statusHow available the issue appears right now: fresh, active, stale, blocked, or waiting on maintainer input.
stale
ClarityHow clearly the issue explains the expected change, acceptance criteria, and next step.
clear
Prerequisites
Pythonpandasnumpyufuncs
Newbie friendlinessA 1-100 score estimating how approachable this issue is for first-time contributors.
65
Research direction
The issue identifies that tests in `test array ufunc.py` generate invalid inputs for ufuncs like arcsin and arccos. To fix, examine the test generation logic in `test ufunc index`, `test ufunc series`, and `test ufunc dataframe` (introduced in PRs #10346 and #10217). Modify input generation to ensure values are within the mathematical domain of each ufunc, while still including some out of domain values to validate consistent error handling. Additionally, compare warnings with pandas to determine if cudf should match them.