P2281R1 Clarifying Range Adaptor Objects

[Casey Carter (CaseyCarter)]

Which explicitly specifies that range adaptor closure objects are perfect-forwarding call wrappers, which are more fastidious about value categories of bound arguments than were our ad-hoc _Partially-applied functions.

Fixes #1983

[Michael Schellenberger Costa (miscco)]

I am missing test for lazy_split_view and split_view which were the views where it does matter

[Casey Carter (CaseyCarter)]

I am missing test for lazy_split_view and split_view which were the views where it does matter

Obviously there are quite a few test cases for each that are still working; what additionally would you like to see tested?

EDIT: I suppose the biggest difference is that lone lvalue arguments to views::split and views::lazy_split are now captured by copy instead of reference, so changes to the original object between capture and evaluation of the range closure won't be visible in the resulting view. Would you like a test for each that validates that difference?

[Stephan T. Lavavej (StephanTLavavej)]

I'm mirroring this to an MSVC-internal PR. Please notify me if any further changes are pushed.

[Stephan T. Lavavej (StephanTLavavej)]

I pushed a workaround for VSO-1409853 "Standard Library Header Units: <ranges> compiler assertion: previous_element == tokenInputStack.TopOfStack(), aliastemplates.cpp 1010".

[Stephan T. Lavavej (StephanTLavavej)]

Thanks for the clarification! 💡 🧠 😹