@vsapsai hi Volodymyr Sapsai , test case clang/test/Modules/hidden-duplicates.m is failing on our llvm community AIX box https://lab.llvm.org/buildbot/#/builders/214/builds/4442/steps/6/logs/FAIL__Clang__hidden-duplicates_m , it gives exit code 70 without detail error message, can you help fix it as soon as possible. Thanks

use std:: to replace _VSTD::

In D118930#3354564, @ldionne wrote:

Please use std:: instead of _VSTD::, though (as suggested by @philnik)

ping for the review. Thanks

please kindly review this patch again. thanks.

Oh, sorry. For some reason I thought z/OS and AIX were the same thing. Are you working on libc++ officially supporting z/OS? If that is the case, maybe it would be a good idea to add a CI node for that?

In D118930#3332943, @philnik wrote:

In D118930#3332884, @NancyWang2222 wrote:

In D118930#3332708, @philnik wrote:

There should also be a test changed somewhere in libcxx/test/std/localization, or is this only part of the fix?

Thanks for providing information, can you provide more detail about what change need for libcxx/test/localization, the patch will fix bunch of Assert d1==d2 issues in std/numerics/rand/rand.dis/rand.dist.bern/ on z/OS related to float point value.

I would have expected test failures in locale.categories/category.numeric/facet.num.get.members/get_float.pass.cpp, since this patch changes behavior of std::num_get on AIX IIUC. So either I'm not understanding something here or there are some test cases missing. Wait there are also no // XFAIL: AIX in std/numerics/rand.dis/rand.dis.bern/. Are these tests currently disabled on AIX somehow?

In D118930#3332708, @philnik wrote:

There should also be a test changed somewhere in libcxx/test/std/localization, or is this only part of the fix?

Can I have 2nd review from libcxx. Thanks.

Can I have 2nd review from libcxx? Thanks

can I have 2nd review from libcxx?

kindly ping for review or approval from libcxx group :)

kindly ping for review or approval from libcxx group

ping :)

@ldionne @Quuxplusone I have addressed comment, Can you help review again. Thanks for the feedback.

@ldionne hi Louis, Can you help review again? Thanks everyone for providing feedback.

change function name letter_value to alphabetical_index

please ignore comment.

change function name and return type

remove extra space

add _LIBCPP_HIDE_FROM_ABI for function __is_valid_in_encoding

match the format suggested by git-clang-format

if no more comments on this PR. plan to land it . Thanks for the feedback.

fix indentation for the function modified.

I rebased with latest master. hope it will fix error.

run git-clang-format for the changes

address comment: change function name from LetterNum to letter_value, and remove extra blank line , also remove constexpr.

fix function name for locale.cpp as well.

change isbasic function name and fix return __c ? true: true statement

ping -:)

ping -:)

ping -:)

include <cctype> so tolower function can be recognized when localization is disabled

need a guard in libcxx/include/charconv as well.

remove an extra curly bracket

put guard in __locale when include locale.h

only include <locale> when _LIBCPP_HAS_NO_LOCALIZATION is not defined

move #include <locale> up one line according to alphabetical order

In D118071#3267542, @mstorsjo wrote:

Sounds reasonable to me - I don't think any other supported OSes would have an issue with that.

Simplify the comments

In D118071#3267521, @yln wrote:

Exact spelling of the temporary test directory shouldn't matter.
The z/OS linker not being able to handle this sounds like a bug... but oh my.

Nit: can we make the comment more compact, maybe:

tmp_dir = tempfile.mkdtemp(prefix='lit-tmp-')  # z/OS linker does not support '_' in paths, so use '-'

LGTM, thanks!

In D107755#2936908, @mstorsjo wrote:

In D107755#2936886, @NancyWang2222 wrote:

hi do you know Why this patch was closed?

I applied it already as it was approved, as I thought you said you can fix any remaining z/OS related details in separate patches later?

hi do you know Why this patch was closed?

hi There is the failing test case libcxx/test/std/input.output/filesystems/class.path/path.member/path.append.pass.cpp , this test case is not in https://reviews.llvm.org/D107124 and https://reviews.llvm.org/D105910 , the failure seems not related to this patch. it fails same without this patch. we will investigate separately. rest of tests in https://reviews.llvm.org/D107124 and https://reviews.llvm.org/D105910 passed. Thanks for the fixes.
Assertion failed: Result == E, file: /plex/wangn/llvm/dev/llvm-project/libcxx/test/std/input.output/filesystems/class.path/path.member/path.append.pass.cpp, line: 250
CEE5207E The signal SIGABRT was received.

The changes look ok to me although I havent tried it on z/OS as long as build pass. I will fix in separate patch if there is an issue on z/OS. just want to confirm those assert wont be run on z/OS and assert macros are empty. I will need some time to download patch and test it out.

Let me know if this is ready, I will need to test it on z/OS.

hi Any suggestion how we should fix those test case, I can add unsupport on z/OS.

@Quuxplusone Thanks Arthur, I can remove delete macro. let me know how windows works if need me to check windows as well.

@Quuxplusone hi Arthur, Can you review ?

Hi I only updated 3 test cases , I didnt modify test cases in https://reviews.llvm.org/D105910 , because @DanielMcIntosh-IBM did some investigation already in those test cases, I will let him handle it. another reason is I am unable to test windows if it requires modify macros for windows as well

Thanks. will look into how ASSERT_VIA_OVERRIDDEN_NEW works

ping :)

@ldionne hi Louis, any question regarding to this PR, I noticed you reviewed similar changes before. Thanks.

In D106151#2890682, @ldionne wrote:

LGTM, but please add short comments (1-2 lines) explaining why the tests are unsupported. I don't need to see this again.

In D106153#2891424, @ldionne wrote:

In D106153#2890692, @ldionne wrote:

LGTM, but please add a 1-2 line comment for each UNSUPPORTED explaining why it fails. You can copy/paste the comment, it's fine. Actually, it even helps cause you can then grep for:

// <the-comment>
// UNSUPPORTED: target={{.+}}-zos{{.*}}

and you'll find all tests that are marked as unsupported for the same reason.

@muiez

Please read the comments even when you get an approval. https://reviews.llvm.org/rG7704fedfff6ef5676adb6415f3be0ac927d1a746 didn't apply my requested change. I LGTM'd the patch because I trusted you'd apply the feedback and didn't want to block you until I could look at it again -- not because no changes were required.

hi Arthur O'Dwyer , I have updated test, can you review it? Thanks

change test cases to be unsupported on z/OS

add unsupported on zOS for 2 failing test cases

hi Arthur O'Dwyer, I am thinking if we can add unsupported cxx-03 for both test cases with current resolution. adding u'some test' or u'a' is right syntax for uchar16_t , same for utf8_t type. any thoughts about this ?

yeah. I also have same question as Sean mentioned. codecvt class for UTF16/32_t starts from c++11, utf8_t started from c++20. we shouldn't run with c++03. the error is expected.

do we know why build are failing ?

In D106151#2883266, @Quuxplusone wrote:

LGTM (if icky) for the char8_t-based tests.
For the other two tests that are currently failing buildkite in C++03 mode (because u's' is a syntax error in C++03), personally I would recommend leaving the code alone and just UNSUPPORTED'ing them on zOS. They're already marked with

// This test runs in C++20, but we have deprecated codecvt<char(16|32), char, mbstate_t> in C++20.

which indicates that it's not too important to make this stuff work on "new" platforms — in fact we're actively deprecating/breaking it on "old" platforms and it might be completely moot by the time C++23 ships.

hi Arthur O'Dwyer , can you kindly review it again? Thanks

ping :)

use new assert pattern so that we can keep old tests