Repository navigation
Introduce the "bare"/"explictly typed" Final terminology - #2370
srittau wants to merge 4 commits into
Conversation
Sorry, something went wrong.
|
|
||
| * A dataclass field may be annotated with ``Final[...]``. For example, ``x: | ||
| Final[int]`` in a dataclass body specifies a dataclass field ``x``, which | ||
| * A dataclass field may be annotated with an :term:`explicitly typed Final` |
There was a problem hiding this comment.
We may have to say what "bare Final" does in dataclasses. I discovered ty handles this buggily: astral-sh/ty#4766.
There are two cases:
- Bare Final with a default. The obvious interpretation is that that gives you a Final dataclass field with a literal type, which is coherent but kind of useless (you can only pass that one specific type).
- Bare Final without a default. Ty and pyrefly implement this like Final[Any], which is sort of coherent but a little surprising to me; mypy disallows it; pyright interprets Final itself as a type which is definitely buggy.
Sorry, something went wrong.
There was a problem hiding this comment.
Oh yes, I failed to mention that: Bare Finals in dataclasses are a thing, but were not mentioned by the specs. I wanted to keep the status quo in this PR, because it's not about that. I think this should be a separate discussion.
Sorry, something went wrong.
jorenham
left a comment
There was a problem hiding this comment.
Thanks!
Sorry, something went wrong.
carljm
left a comment
There was a problem hiding this comment.
Thanks!
Sorry, something went wrong.
Co-authored-by: Joren Hammudoglu <jhammudoglu@gmail.com>
The main changes:
The rest of the changes use the new terms and link to the new glossary entries.
Closes: #2356