Maybe it's worth cleaning up __decay_copy in this patch too? It's probably best if we commit separately to D100225.

updates synopsis, rebases to activate CI

rebases to activate CI

rebases to activate CI

updates synopsis, rebases to activate CI

updates synopsis, rebases to activate CI

updates synopsis, rebases to activate CI

adds synopsis, rebases to activate CI

rebases to activate CI

rebases to activate CI

rebases to activate CI

rebases to activate CI

rebases to activate CI

rebases to activate CI

addresses feedback from @Quuxplusone and @tcanens

re-adds missing underscore that was deleted in clang-format cleanup

updates commit message (NFC)

corrects and suppresses clang-format due to its inexperience with concepts

fixes dependency in commit message (NFC)

corrects typo tanking CI

fixes iter_move per @tcanens' feedback
moves __class_or_enum out of __swap scope since it's useful elsewhere

moves iterator_traits into __iterator_traits as per @tcanens' feedback
hopefully gets CI working

applies LWG3420 and (hopefully) fixes locale build failure

renames all the cryptic types imported from cjdb-ranges

moves types common to indirectly_readable and indirectly_writable tests into a header

rebases to activate CI

replaces conditional_t with _If

rebases to activate CI

rebases to activate CI

fixes GCC compile-time error with a workaround

rebases to activate CI, applies @curdeius' comment suggestions from D96477 (putting them in this patch makes more sense due to this patch's detachment from anything specific)

rebases to activate CI (hopefully for the last time)

adds guards around the filesystem stuff

adds pragma to test to silence GCC warning

In D99141#2669722, @ldionne wrote:

I'd like to put an end to comments about "too many tests" once and for all. I think we all agree that test coverage is a good property to have. I think we also agree that readability is a good property to have, and that code duplication is bad. We want both, as much as possible.

I think what we're seeing here is an example where we have a lot of code coverage, which is great, however the cost (right now) is a lot of repetition, and hence decreased readability. For me, the solution is not to remove the test coverage, but instead to implement it without reducing readability so much. Here's an example of how this can be done (there are other ways, for example we could use templates instead but I just hacked this together). This replaces lines 174 to 702. Yes it's still not *small*, but it's much better and IMO easier to read, while not dropping any of the coverage. And it would be easy to make it even better by using two nested "loops" over all qualifiers, but it's unclear to me that would actually increase readability.

...

WDYT? If we have to do this sort of thing frequently, we could also bundle a few utilities under support/ to reduce duplication further.

applies @ldionne's proposed test framework

rebases to fix CI

@zoecarver @curdeius this patch now changes a feature-test macro, so I don't think it can be NFC. Are you okay with that?

properly updates feature-test macro per @Mordante's request

adds assert and makes commit message more descriptive

corrects self-subsumption test

This patch has been open for almost two months now, so if there's nothing else that urgently needs attention, I'd like to see it merged today.

changes incorrect version from 14.0 to 13.0 and renames incorrectly-named overload

rebases to activate CI

rebases to activate CI

rebases to activate CI

rebases to activate CI

adds internal ranges feature-test macro

rebases to activate CI

makes iter_difference_t and iter_value_t SFINAE-friendly and consolidates test files

rebases to activate CI

applies @Mordante's feedback

applies feedback from @miscco and @Mordante

enables feature-test macro, revises commit message

gets CI passing

removes zero-width joiners

rebases to activate CI

fixes include guard to guard on concepts

applies @zoecarver and @curdeius' feedback