Handle annotated indexes #762
Conversation
| @dataclass(frozen=True) | ||
| class IndexedAnnotation: | ||
| _indexed: Tuple[int, Dict[str, Any]] | ||
|
|
||
|
|
||
| def Indexed(typ=None, index_type=ASCENDING, **kwargs): |
There was a problem hiding this comment.
If we don't want to keep the same interface for both ways of defining indexes, we could just have the users instantiate IndexedAnnotation when using Annotated. Though in this case I would clean up the interface for this class and probably the name for a better one.
Maybe we could even add a deprecation warning to Indexed if we intend to prefer the Annotated route.
There was a problem hiding this comment.
Maybe the name could be IndexedField or IdxField?
There was a problem hiding this comment.
My 2¢:
-1 on introducing a new name, Indexed is fine.
+1 for emitting a deprecation warning if typ is not None (at least for Python 3.8+/Pydantic v2; not sure if Annotated is supported for Pydantic v1).
There was a problem hiding this comment.
In pydantic v1, annotated is supported. It is even used for discriminating unions
There was a problem hiding this comment.
If we don't want to keep the same interface for both ways of defining indexes
It is better to keep both interfaces. Otherwise this breaking change will affect too many users.
There was a problem hiding this comment.
Yeah, I meant keeping the old one, but making a different one for the new way. Anyways, that's your call, I'm glad you liked it!
|
Thank you for the PR! Merged. It will be published tomorrow |
|
Is this applied to Bunnet as well? |
This PR aims to introduce a way to declare indexed fields using python's
Annotated, extending the current use ofIndexed. It implements the syntax proposed in #638.It also supports the use of
UUIDandEmailStrwhen using Pydantic 2 andAnnotated. Kinda solving #701, though I couldn't reproduce the issue withPydanticObjectId.