Question on Type Annotation and Docstring about pyglove HOT 5 CLOSED

PyGlove only supports the latter syntax. For the former one, the field_a string is not associated with the field.

from pyglove.

Yeah, it seems that the latter syntax is the official way of attribute annotation. Because the second args can go far beyond the string type, it was originally designed to attach some metadata in the annotation.

from pyglove.

I know :-) In my opinion Annotated is more scalable in providing extra information for a field, based on which we can easily support value spec & field metadata in near future. These features are already part of pg.members.

We may consider support the "official" attribute annotation if our fellow users prefer this way, although I personally don't like it for its poorer readability and scalability. But let me know if you are in a urgent need.

from pyglove.

lol, no urgent need. The external IDEs do not view Annotated[.., x]'s s as annotation and thus will ignore them, just a heads up on the incompatibility of those common softwares vs. the practice in PG~

from pyglove.

Good to know.

It seems the "former" form of field annotation (a new line of string following the field definition) was proposed in PEP-224 but got rejected. So maybe we should wait a bit see the necessity of this support.

from pyglove.