Typical

Latest version: v2.8.1

Safety actively analyzes 629564 Python packages for vulnerabilities to keep your Python projects secure.

Scan your dependencies

Page 3 of 12

2.7.1

Enhancements
- `typical` will now default to [orjson](https://github.com/ijl/orjson) for json serialization and deserialization if it is installed. This provides up to 3X performance boost, but has different behavior and output than ujson or stdlib json.
- Callable class instances are now treated as function types when wrapped (thank you xbanke, 173)

Fixes
- Expose the `always` parameter to `typic.klass`, since we are raising warnings about its usage (177)

2.7.0

Features

- Wrapped routines and frozen dataclasses now coerce inputs up to 10x faster than before (on-par with wrapped classes or bound protocols) (175).
- More descriptive and correct type-hints for the public API.
- Type-hinting with an abstract base class (ABC) will now result in validation against that type rather than coercion (170)

Fixes

- Resolves issue where we'd sometimes fail to clean up the repr of subscripted generics (170)
- Resolves issue where we failed to locate the correct target type for a value within a Union of overlapping types (170)
- Resolves issue where we'd iterate over the fields of a non-iterable, primitive type when coercing to a collection, rather than raise a TypeError as expected (174).

2.6.4

2.6.3

Fixes

- Add intelligent instance-checks for unions (resolves 167)
- Make `field.default` positional (resolves 168)
- Add `__version__` to `typic.__init__` and add `bumpver` to manage multiple versions (resolves 168)

2.6.2

Fixes

- This release resolves an issue when building serializers for Unions of pandas collection types (e.g., DataFrame, Series) which prevented proper compilation (166)

Changes

- Tagged Unions must be tagged with "public" fields (e.g., no fields starting with `_` are considered).

2.6.1

Fixes
- Fixes dynamic evaluation of serializers for unknown types within containers.

Page 3 of 12

© 2024 Safety CLI Cybersecurity Inc. All Rights Reserved.