An Empirical Study of Type-Related Defects in Python Projects
Bibliographic record
Abstract
In recent years,Pythonhas experienced an explosive growth in adoption, particularly among open source projects. WhilePython's dynamically-typed nature provides developers with powerful programming abstractions, that same dynamic type system allows for type-related defects to accumulate in code bases. To aid in the early detection of type-related defects, type annotations were introduced into thePythonecosystem (i.e., PEP-484) and static type checkers likemypyhave appeared on the market. While applying a type checker likemypycan in theory help to catch type-related defects before they impact users, little is known about the real impact of adopting a type checker to reveal defects inPythonprojects. In this paper, we study the extent to whichPythonprojects benefit from such type checking features. For this purpose, we mine the issue tracking and version control repositories of 210Pythonprojects on GitHub. Inspired by the work of Gaoet al.on type-related defects in JavaScript, we add type annotations to test whethermypydetects an error that would have helped developers to avoid real defects. We observe that 15 percent of the defects could have been prevented bymypy. Moreover, we find that there is no significant difference between the experience level of developers committing type-related defects and the experience of developers committing defects that are not type-related. In addition, a manual analysis of the anti-patterns that most commonly lead to type-checking faults reveals that the redefinition ofPythonreferences, dynamic attribute initialization and incorrectly handled Null objects are the most common causes of type-related faults. Since our study is conducted on fixed public defects that have gone through code reviews and multiple test cycles, these results represent a lower bound on the benefits of adopting a type checker. Therefore, we recommend incorporating a static type checker likemypyinto the development workflow, as not only will it prevent type-related defects but also mitigate certain anti-patterns during development.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.035 | 0.319 |
| Meta-epidemiology (narrow) | 0.000 | 0.001 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.006 | 0.005 |
| Science and technology studies | 0.002 | 0.003 |
| Scholarly communication | 0.003 | 0.006 |
| Open science | 0.002 | 0.003 |
| Research integrity | 0.002 | 0.003 |
| Insufficient payload (model declined to judge) | 0.003 | 0.001 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".