Regarding the title, there seem to be two remaining issues.
@no_type_check isn't implemented for classes
@no_type_check_decorator isn't implemented at all
Note that the latter allows people to define their own decorators whose effect is equivalent to that of @no_type_check, i.e. this is a meta-level decorator.
AFAICT, #6584 will fix the first bullet but not the second. Perhaps we need to create a separate issue for the latter?
OK, changed the title to reflect that this issue (#607) is about the @no_type_check decorator as applied to classes, while the new #6583 is about the "meta" decorator, @no_type_check_decorator.
There is little evidence for any user demand for this feature, so this is overall unlikely to be implemented unless you, the reader, are willing to do it.
FWIW re: user demand, I ran into wanting this feature which is why I'm subscribed. I wanted to be able to use the type annotation syntax for defining structs for generated code for IPC, but the types I would use wouldn't make sense to a normal type checker, and the class is only intended to be used for code generation via inspect, not actually instantiated. I think there is a chicken and egg problem where people aren't going to demonstrate use cases as long as it doesn't work. In my case I abandoned the idea because I couldn't get any of the static type checkers to cooperate and instead used kwarg syntax and rely on the fact that in newer python versions kwarg order is preserved, e.g.
Activity
JukkaL commented on May 17, 2015
This is now supported for functions but not for classes.
lunixbochs commented on Mar 22, 2019
The title here is incorrect, the @no_type_check decorator is working for functions but @no_type_check_decorator does not appear to be implemented.
gvanrossum commented on Mar 22, 2019
Regarding the title, there seem to be two remaining issues.
Note that the latter allows people to define their own decorators whose effect is equivalent to that of @no_type_check, i.e. this is a meta-level decorator.
AFAICT, #6584 will fix the first bullet but not the second. Perhaps we need to create a separate issue for the latter?
gvanrossum commented on Mar 22, 2019
OK, changed the title to reflect that this issue (#607) is about the @no_type_check decorator as applied to classes, while the new #6583 is about the "meta" decorator, @no_type_check_decorator.
lunixbochs commented on Jan 21, 2020
Can someone review my PR #6584?
hauntsaninja commented on Feb 4, 2024
Linking a thread about what the semantics for no_type_check for classes should be: https://discuss.python.org/t/no-type-check-decorator/43119
There is little evidence for any user demand for this feature, so this is overall unlikely to be implemented unless you, the reader, are willing to do it.
Jelle mentions a possible semantics here: https://discuss.python.org/t/no-type-check-decorator/43119/34
Another possible semantics is basically replacing the class with class X(Any): ...
jgarvin commented on Mar 12, 2024
FWIW re: user demand, I ran into wanting this feature which is why I'm subscribed. I wanted to be able to use the type annotation syntax for defining structs for generated code for IPC, but the types I would use wouldn't make sense to a normal type checker, and the class is only intended to be used for code generation via inspect, not actually instantiated. I think there is a chicken and egg problem where people aren't going to demonstrate use cases as long as it doesn't work. In my case I abandoned the idea because I couldn't get any of the static type checkers to cooperate and instead used kwarg syntax and rely on the fact that in newer python versions kwarg order is preserved, e.g.
mystruct = gen.Struct( x=gen.Int32, y=gen.Float64, # ... )