Clarify that tuple should not be prohibited as an argument to type#2146
Clarify that tuple should not be prohibited as an argument to type#2146carljm merged 2 commits intopython:mainfrom
tuple should not be prohibited as an argument to type#2146Conversation
Closes python#2145 This spec change reconciles the typing spec and the conformance testsuite. `tuple` is an object of the `type` class, so it should be accepted as a value of `type`.
srittau
added
Typing Council decision
erictraut
left a comment
erictraut
left a comment
There was a problem hiding this comment.
I don't think the new wording should mention tuple at all. It's not a special form. It's simply a class.
The problem here is the phrase special constructs, which is not defined anywhere in the spec and is therefore ambiguous. I think we should change this to special form, which is defined in the glossary.
fangyi-zhou
changed the title
tuple should be accepted as an argument to typetuple should not be prohibited as an argument to type
|
Maybe add an extra sentence now? |
No, this is already covered elsewhere in the spec. We try to avoid redundancies because they create confusion and potential inconsistencies. |
@erictraut Do you mean this is covered in the grammar in https://typing.python.org/en/latest/spec/annotations.html#type-and-annotation-expressions? |
Yes |
rchen152
left a comment
rchen152
left a comment
There was a problem hiding this comment.
I agree with Jelle and Eric's points at https://discuss.python.org/t/spec-change-clarify-that-tuple-should-not-be-prohibited-as-an-argument-to-type/105590 that this section is in need of a broader fix, but this change as-is is already an improvement (removes the obviously incorrect statement that type[tuple] is not allowed and the ambiguous term "special constructs"), so I'm in favor of accepting it.
carljm
left a comment
carljm
left a comment
There was a problem hiding this comment.
Agreed that is is already an improvement, even if we should also make a more comprehensive fix.
|
Are we happy to merge the current change as is? Or we would want to take this opportunity to improve upon it more? |
|
I am happy to merge this as-is and consider broader improvements separately, but I would like more than two typing council members to approve before I merge it. |
|
I'm good with this change. |
6 participants
Closes #2145
This spec change reconciles the typing spec and the conformance testsuite.
tupleis an object of thetypeclass, so it should be accepted as a value oftype.