Various improvements of test suite before addition of `<flat_meow>`. #5968

vmichal · Dec 19, 2025

This PR is a collection of changes that are remotely related to work on <flat_meow>, but they are sufficiently self-contained that it would be nice to introduce them separately, before the feature branch is merged.

Changes:

Add tests LWG3987_including_flat_foo_doesnt_provide_begin_end for requirements of [iterator.range]/1 that certain headers introduce free functions begin/end/data/size etc. This test is required to prove proper implementation of LWG-3987 "Including <flat_foo> doesn't provide std::begin/end". Structure inspired by GH_000545_include_compare. Some notes:
- <valarray> test is deactivated with // TRANSITION as it is blocked by Implement P3016R6 Resolve Inconsistencies In begin/end For valarray And Braced Initializer Lists #5847 (expected to accept soon)
- For completeness of the test, I have already added <flat_map>, <flat_set>, <optional>, <hive> and <inplace_vector> tests guarded by corresponding feature test macros; they will be activated by resolving P0429R9 <flat_map> #2910, P1222R4 <flat_set> #2912, P3168R2 std::optional Range Support #4772, P0447R28 <hive> #5301 and P0843R14 <inplace_vector> #4766, respectively.

… feature test macro.

…cros and _HAS_CXX2a

frederick-vs-ja · Dec 19, 2025

tests/std/tests/LWG3987_including_flat_foo_doesnt_provide_begin_end/test_inplace_vector.cpp

@@ -0,0 +1,20 @@
+// Copyright (c) Microsoft Corporation.


I think we should drop this file at this moment. It should be added when implementing inplace_vector.

Although I have created a soft link by mentioning the relevant issues, I think relying on authors of inplace_vector, hive or optional range accessors to update this test in the future is a completely redundant mental burden and waste of time.

I like the current solution since it tests the current WP's requirements completely and will continue to do so until a new container is added. I personally see no issue in testing "future technology" when it is properly guarded with feature test macros. I consider this in line with their intended use case. Especially in this case, where I don't speculatively test some unstable API.

frederick-vs-ja · Dec 19, 2025

tests/std/tests/LWG3987_including_flat_foo_doesnt_provide_begin_end/test_hive.cpp

+#include <version>
+#if defined(__cpp_lib_hive) && _HAS_CXX26
+
+#include <hive>


Ditto, this file should be added when implementing hive.

frederick-vs-ja · Dec 19, 2025

tests/std/test.lst

 tests\LWG3545_pointer_traits_sfinae
 tests\LWG3561_discard_block_engine_counter
 tests\LWG3610_iota_view_size_and_integer_class
+tests\LWG3987_including_flat_foo_doesnt_provide_begin_end


I don't think LWG3987_including_flat_foo_doesnt_provide_begin_end is a good name for these tests. As availability of begin/end in many headers is required since C++11 via WG21-N2930.
But since N2930 is too old, I guess it's a bit unconventional to mention it in the test name.

Also, we generally don't mention "foo" in the repo, even if the LWG issue title says it.

Also, we generally don't mention "foo" in the repo, even if the LWG issue title says it.

I don't mean to disagree, but can you elaborate a little? Yes, I have noticed you prefer meow over foo, but as far as I am aware, both are perfectly valid placeholders with no cultural, religious or any different negative connotation.

I don't think LWG3987_including_flat_foo_doesnt_provide_begin_end is a good name for these tests. As availability of begin/end in many headers is required since C++11 via WG21-N2930. But since N2930 is too old, I guess it's a bit unconventional to mention it in the test name.

Oh yes, I agree, but I found no precedent for naming a test that is not strictly related to any paper (Pxxx, LWGxxx) or issue (GH, and I assume VSO and Dev fall into this category too).
I'll gladly rename this to (some-prefix)_include_iterator_range, but regarding "some-prefix", I would need to use this PR as GH_005968 or created a completely new prefix, e.g. MISC for unsorted miscellaneous tests.

I prefer the MISC prefix, but it is your call.

There was an old policy on MSDN where they avoided "foo" because it was part of "foobar" and people were concerned that sounded like something else. Really silly but we've continued to avoid "foo".

We used to use VSO_0000000 for novel tests but that was clutter-inducing so we stopped. Marking this as GH with the PR number is useful context, I recommend that.

frederick-vs-ja · Dec 19, 2025

tests/std/tests/LWG3987_including_flat_foo_doesnt_provide_begin_end/test.cpp

+// Copyright (c) Microsoft Corporation.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+
+#include <version>


I think we should either avoid including any header, or include <version> and then guard tests with _HAS_CXXNN.

frederick-vs-ja · Dec 19, 2025

tests/std/tests/LWG3987_including_flat_foo_doesnt_provide_begin_end/test.cpp

+void test_forward_list();
+void test_hive();
+void test_inplace_vector();
+void test_list();


test_iterator() seems missing.

frederick-vs-ja · Dec 19, 2025

tests/std/tests/LWG3987_including_flat_foo_doesnt_provide_begin_end/test_optional.cpp

+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+
+#include <version>
+#if defined(__cpp_lib_optional_range_support) && _HAS_CXX17


I believe __cpp_lib_optional_range_support won't be defined in pre-C++26 modes. Perhaps we should also drop this file at this moment.

I believe __cpp_lib_optional_range_support won't be defined in pre-C++26 modes. Perhaps we should also drop this file at this moment.

I only skimmed through yvals_core.h and - clearly incorrectly - concluded that we define all __cpp_lib_xxx unconditionally as soon as a feature is implemented. So I will drop all _HAS_CXX_2a and keep just the feature test macro.

StephanTLavavej · Dec 19, 2025

Draft mode just disables CI runs. Was that your intention?

vmichal · Dec 19, 2025

Draft mode just disables CI runs. Was that your intention?

Today I learned :D
My intention was to indicate that this is probably not the final state of this PR and thus should not be merged yet (in case you were extra eager to work during vacation). I am working in parallel (although slowly) on changes to the flat_meow feature branch and this, so more tests may be added to this PR.

StephanTLavavej · Dec 19, 2025

Thanks, I'll move it to WIP for you. When you're ready to run tests, mark it ready for review (moving it out of draft status) and then push commits, which will trigger CI. Merely moving out of draft mode does not start tests.

vmichal added 5 commits December 19, 2025 01:25

Test that selected headers introduce machinery from iterator.range/1

683ef99

Properly guard the test of introduction of begin/end by <optional> by…

c7c0490

… feature test macro.

Add tests for hive and inplace_vector.

0d52345

Use feature test macro instead of __has_include.

3e32558

Properly guard tests for C++23 and C++26 feature with feature test ma…

a48d2e0

…cros and _HAS_CXX2a

github-project-automation bot added this to STL Code Reviews Dec 19, 2025

github-project-automation bot moved this to Initial Review in STL Code Reviews Dec 19, 2025

frederick-vs-ja reviewed Dec 19, 2025

View reviewed changes

StephanTLavavej added the test Related to test code label Dec 19, 2025

vmichal added 3 commits December 19, 2025 13:04

Properly test the free functions, not member functions.

8c92460

Use only feature test macros, do not duplicate the check with _HAS_CXX2a

53454ae

Add test for <iterator>

dc2cd24

StephanTLavavej moved this from Initial Review to Work In Progress in STL Code Reviews Dec 19, 2025

Search code, repositories, users, issues, pull requests...

Various improvements of test suite before addition of <flat_meow>. #5968

Are you sure you want to change the base?

Various improvements of test suite before addition of <flat_meow>. #5968

Uh oh!

Conversation

vmichal commented Dec 19, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

vmichal Dec 19, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

frederick-vs-ja Dec 19, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

StephanTLavavej commented Dec 19, 2025

Uh oh!

vmichal commented Dec 19, 2025

Uh oh!

StephanTLavavej commented Dec 19, 2025

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

Various improvements of test suite before addition of `<flat_meow>`. #5968

Various improvements of test suite before addition of `<flat_meow>`. #5968

vmichal commented Dec 19, 2025 •

edited

Loading

vmichal Dec 19, 2025 •

edited

Loading

frederick-vs-ja Dec 19, 2025 •

edited

Loading