際際滷

際際滷Share a Scribd company logo

Type Checking in Python at Tiqets

scar Vilaplana
scar Vilaplana

How we use Python type checking at Tiqets and what we love about it. Read more at http://tiqets.engineering

1 of 27
Download to read offline
How we use it and what we love about it Type checking in Python
Making culture more accessible
Why do we use Python type checking?
Better Developer Experience No guessing & no surprises Less cognitive load Type checking at commit & build time Realize mistakes early Type checks are easier to write & maintain Avoid trivial unit tests Type checking at runtime with Pydantic Data validation
Type Checking in Python at Tiqets
What types can I use? Collections Tuple, Dict, MutableMapping, List, NamedTuple... Composers Union, Optional Callable Generics TypeVar, Generic, Protocol ...and all simple types int, str, bytes...
Checks: Green No more typing issues Add typing to new code Add typing on existing code How did we introduce type checks? Make the changes needed to have mypy run without errors. Add a mypy step to your CI build and pre-commit hooks. From now onwards, no new typing problems can be introduced. Things can only get better! Teach your team about typing and mypy. Agree on adding type hints to all new code. Gradually add type hints to all existing code. If you touch it, you add types.
Pitfalls False sense of security It wont catch all issues. No runtime checks. Type checking: an extra safety step, not a replacement. No optimizations Python wont use the knowledge about types to make any optimizations on your code. Untyped libraries will lead to typing mistakes. Type hints can get ugly Tip on that is: if your type is ugly, is there a better approach?
Beyond the basics
Type Checking in Python at Tiqets
Type Checking in Python at Tiqets
Type Checking in Python at Tiqets
Type Checking in Python at Tiqets
Type Checking in Python at Tiqets
A 鍖nal class cant be inherited from. A 鍖nal method cant be overloaded. A 鍖nal variable cant be reassigned.
Trickier types
Type Checking in Python at Tiqets
Type Checking in Python at Tiqets
Type Checking in Python at Tiqets
Type Checking in Python at Tiqets
Type Checking in Python at Tiqets
Type Checking in Python at Tiqets
Type Checking in Python at Tiqets
Type checkers Reference implementation. We use it at Tiqets mypy Faster; runs on nodejs. pyright (Microsoft) Can infer types from unannotated code. pytype (Google) Can do incremental checking pyre (Facebook)
Whats coming up? Uni鍖ed error codes Easier integration of type checkers with other tools Less CamelCase list[dict] instead of List[Dict] Shorter syntax int | float instead of Union[int, float] ?str instead of Optional[str]
Check out our blog at tiqets.engineering Thanks!
Type Checking in Python at Tiqets

More Related Content

Type Checking in Python at Tiqets

  • 1. How we use it and what we love about it Type checking in Python
  • 3. Why do we use Python type checking?
  • 4. Better Developer Experience No guessing & no surprises Less cognitive load Type checking at commit & build time Realize mistakes early Type checks are easier to write & maintain Avoid trivial unit tests Type checking at runtime with Pydantic Data validation
  • 6. What types can I use? Collections Tuple, Dict, MutableMapping, List, NamedTuple... Composers Union, Optional Callable Generics TypeVar, Generic, Protocol ...and all simple types int, str, bytes...
  • 7. Checks: Green No more typing issues Add typing to new code Add typing on existing code How did we introduce type checks? Make the changes needed to have mypy run without errors. Add a mypy step to your CI build and pre-commit hooks. From now onwards, no new typing problems can be introduced. Things can only get better! Teach your team about typing and mypy. Agree on adding type hints to all new code. Gradually add type hints to all existing code. If you touch it, you add types.
  • 8. Pitfalls False sense of security It wont catch all issues. No runtime checks. Type checking: an extra safety step, not a replacement. No optimizations Python wont use the knowledge about types to make any optimizations on your code. Untyped libraries will lead to typing mistakes. Type hints can get ugly Tip on that is: if your type is ugly, is there a better approach?
  • 15. A 鍖nal class cant be inherited from. A 鍖nal method cant be overloaded. A 鍖nal variable cant be reassigned.
  • 24. Type checkers Reference implementation. We use it at Tiqets mypy Faster; runs on nodejs. pyright (Microsoft) Can infer types from unannotated code. pytype (Google) Can do incremental checking pyre (Facebook)
  • 25. Whats coming up? Uni鍖ed error codes Easier integration of type checkers with other tools Less CamelCase list[dict] instead of List[Dict] Shorter syntax int | float instead of Union[int, float] ?str instead of Optional[str]
  • 26. Check out our blog at tiqets.engineering Thanks!