Thanks for the review, could somebody commit it as I do not have the rights

Thanks for the review, could somebody commit it as I do not have the rights

Thanks for the review, could somebody commit it as I do not have the rights

Thanks for the review. I do not have commit rights though.

Only update spans feature test macro

It seems it finally worked. No idea what went wrong as the changes were still in there

second try

Add changes?!

That is unfortunate, i will restore them later in the evening

Rebase on master

@ldionne Thanks a lot for the hint. I updated the generate_feature_test_macro_components.py script and also fixed the incorrect value provided to __cpp_lib_to_array

• [libcxx][span] Update synopsis
• Update feature test macro

• [libcxx][span] Update synopsis

@ldionne Thanks a lot for the review. I have added the feature test macro as I did not find it anywhere else.

ping

ping

ping

ping

Ping

In D75642#1995120, @zoecarver wrote:

Looks like there are still white-space changes in the tests. And there are warning from the CI? Are those formatting warnings or "real" warnings?

ping, this is now without whitespace changes so I would guess it is ready to go?

I left mostly some nits with respect to readability. Great job

Thanks to @CaseyCarter for the example

Add test for a reference wrapper type

Actually I was having trouble comming up with a test case for bullet 4.

ping

ping

ping

@ldionne do you require any changes here (especially regarding the feature test macro)

Ping, this is also all red

Ping, this is all red so...

Ping, I guess the clang format should not be applied?

[libcxx][span] Update synopsis

Thanks for the review, I am unsure whether one should really update the feature test macro without the other changes in place (tuple interface, relaxed array constructors etc)

span-ing more pings

ping

gentle ping

Do not require concepts

You are absolutely right.

gentle ping

Do not clang-format the code

My hope was that it would fix the complaints from the Harbormaster. But that doesnt seem to be the case

clang-format all the files

Fix whitespaces

Ah sorry about that, I overlooked that we had special cased tests without count.

That is indeed the correct thing to do +1

I believe this is superseded by the implementation of P1976R2 in D74577

Add failing tests for explicit constructors

Forgot the explicit for the container constructors

[libcxx][span] Use requires clause for default constructor

Some nits, looks great, especially for library concepts. Pinging @cjdb

rebased

rebased

Make STL happy, as they apply [[nodiscard]] everywhere

Remove superfluous newlines

Add tests

@ldionne I aggree with your sentiment and I only discovered this when I implemented span for MSVC STL and added the same static assert to operator[]

Simplified code and improved comments

@cjdb I have to disagree. That would create a ciruclar dependency between concepts and type_traits, which is in my opinion not worth it.

For what its worth, I found some time today to cleanup what I had already done with @CaseyCarter old implementation D49118 (common_type and common_reference were missing).

This looks already quite good.

Comment about empty arrays

Ping as this is an actual bugfix

Gentle Ping

Gentle Ping

Ping trivially correct change

Gentle ping with the relevant quote from the standard

• [libcxx] span: Check default constructible

Added missing include of <version>

Added type_traits and concepts check for the different spans

• [libcxx] span: Check default constructible

clang-format comment

Library concepts are not yet implemented.

Removed the forward declaration of array as that would be far more involved.

Do not forward declare array

Hm, this fixes a todo that was written by you..

Some comments

Updating D71999: [libcxx][tests] span: Add failing tests for span::first and span::last

Updating D71999: [libcxx][tests] span: Add failing tests for span::first and span::last