Update to Clang 10 and Visual Studio 2019 version 16.6p2 by BillyONeal · Pull Request #645 · microsoft/STL

BillyONeal · 2020-03-27T07:49:44Z

re-clang-formats the tree to comply with new clang 10 clang-format
updates our agents to F series VMs which are faster and cheaper for our build workloads
defaults scale set to 0 VMs and lets the Azure Pipelines service control all resizing
fix tests to pass with the new compilers
adds antivirus exclusions for the build directory

Co-authored by Casey Carter and Curtis Bezault

miscco · 2020-03-27T08:22:21Z

So for me I assume that these are exclusively formatting changes in the other files and all functional changes between clang-9 / clang-10 will only come at a later stage?

StephanTLavavej

SO AMAZING

StephanTLavavej · 2020-03-27T08:37:03Z

stl/inc/functional

        & noexcept(noexcept( //
            _Call_front_binder(_Seq{}, _Mypair._Get_first(), _Mypair._Myval2, _STD forward<_Unbound>(_Unbargs)...))) //
-          -> decltype( //
-              _Call_front_binder(
-                  _Seq{}, _Mypair._Get_first(), _Mypair._Myval2, _STD forward<_Unbound>(_Unbargs)...)) { //
+        -> decltype( //
+            _Call_front_binder(_Seq{}, _Mypair._Get_first(), _Mypair._Myval2, _STD forward<_Unbound>(_Unbargs)...)) { //


Empty comments were being used here to improve clang-format's behavior. If you remove them, does clang-format 10 do something reasonable-looking now?

I admittedly didn't even read what clang-format did :P

That's what I'm for, to read the output of humans and robots alike, line by repetitive line, until I find the single wrong one.

StephanTLavavej · 2020-03-27T08:39:12Z

stl/inc/random

+    } else if _CONSTEXPR_IF (_Cx <= UINT_MAX
+                             && static_cast<_Uint>(_Mx - 1)
+                                    <= (UINT_MAX - _Cx)
+                                           / _Ax) { // unsigned int is sufficient to store intermediate calculation


Now that _CONSTEXPR_IF isn't horribly damaging this formatting, it appears that this comment should be detached to improve readability.

StephanTLavavej · 2020-03-27T08:39:25Z

stl/inc/random

+                                    <= (ULLONG_MAX - _Cx) / _Ax) { // unsigned long long is sufficient to store
+                                                                   // intermediate calculation


Also detach comment?

StephanTLavavej · 2020-03-27T08:40:15Z

stl/inc/system_error

-_NODISCARD inline error_condition error_category::default_error_condition(int _Errval) const
-    noexcept { // make error_condition for error code
+_NODISCARD inline error_condition error_category::default_error_condition(
+    int _Errval) const noexcept { // make error_condition for error code


This comment should be detached according to our modern conventions.

StephanTLavavej · 2020-03-27T08:40:28Z

stl/inc/system_error

-_NODISCARD inline error_condition error_code::default_error_condition() const
-    noexcept { // make error_condition for error code
+_NODISCARD inline error_condition
+    error_code::default_error_condition() const noexcept { // make error_condition for error code


Also detach.

StephanTLavavej · 2020-03-27T08:41:12Z

stl/inc/type_traits

 template <class _Ty1, class _Ty2>
-struct _Swap_cannot_throw : bool_constant<noexcept(swap(_STD declval<_Ty1>(), _STD declval<_Ty2>()))
-                                          && noexcept(swap(_STD declval<_Ty2>(), _STD declval<_Ty1>()))> {
+struct _Swap_cannot_throw : bool_constant<noexcept(swap(_STD declval<_Ty1>(), _STD declval<_Ty2>()))&& noexcept(


Ugh, this is a regression in appearance. Add an empty comment to force wrapping before the logical AND?

StephanTLavavej · 2020-03-27T08:41:51Z

stl/inc/variant

-    _NODISCARD constexpr size_t index() const
-        noexcept { // index of the contained alternative or variant_npos if valueless_by_exception
+    _NODISCARD constexpr size_t
+        index() const noexcept { // index of the contained alternative or variant_npos if valueless_by_exception


Detach comment.

BillyONeal · 2020-03-27T09:03:27Z

So for me I assume that these are exclusively formatting changes in the other files and all functional changes between clang-9 / clang-10 will only come at a later stage?

Correct, except I'm going to have to make 2 functional changes to tests to make them pass.

tests/std/tests/P0513R0_poisoning_the_hash/test.cpp

CaseyCarter · 2020-03-27T16:28:07Z

Correct, except I'm going to have to make 2 functional changes to tests to make them pass.

Fixes for the feature-test macro test are ~~here~~ in this branch. I'm not sure what's going on with the spaceship test; it looks like cl runs are "failing" with exit status 0?

cbezault · 2020-03-27T16:36:39Z

They're XPASSing. It looks like there's improvement needed in the errors I output from the test harness in that case.

CaseyCarter · 2020-03-27T16:47:10Z

They're XPASSing. It looks like there's improvement needed in the errors I output from the test harness in that case.

Ah: C:\Program Files (x86)\Microsoft Visual Studio\2019\Preview\VC\Tools\MSVC\14.26.28720\bin\HostX64\x86\cl.EXE is Preview 2 - no wonder.

tests/std/expected_results.txt

CMakeLists.txt

* re-clang-formats the tree to comply with new clang 10 clang-format * updates our agents to F series VMs which are faster and cheaper for our build workloads * defaults scale set to 0 VMs and lets the Azure Pipelines service control all resizing * fix tests to pass with the new compilers Co-authored by Casey Carter and Curtis Bezault

README.md

azure-devops/create-new-agent-image.ps1

StephanTLavavej · 2020-03-29T01:57:23Z

I observe that it's difficult to review changes when you squash away history 😿

StephanTLavavej · 2020-03-29T02:01:45Z

Fortunately everything looks good (GitHub at least remembers which files have been Viewed and are unchanged). Thanks.

BillyONeal · 2020-03-29T02:14:58Z

I observe that it's difficult to review changes when you squash away history 😿

Sorry, without squashing it was getting awkward to apply to the msvc repo :/

StephanTLavavej · 2020-03-29T02:17:27Z

I'm not sure how you port changes from GitHub to MSVC, but I've found that I can port arbitrarily complex histories to MSVC. As long as the GitHub branch is either rebased or merged with the latest master, a simple git diff master WHATEVER_BRANCH > meow.patch is sufficient to create an inherently squashed patch to apply to MSVC.

This was skipped by microsoftGH-682 on 2020-04-09. At the time, we were using VS 2019 16.6 Preview 2 (microsoftGH-645 merged on 2020-03-28). Now we're using VS 2019 16.8 Preview 1, and I can't reproduce this ICE locally. I ran x86 and x64, 30 times each, and they all passed. Let's try restoring this configuration.

BillyONeal requested a review from a team as a code owner March 27, 2020 07:49

BillyONeal self-assigned this Mar 27, 2020

BillyONeal added the infrastructure Related to repository automation label Mar 27, 2020

StephanTLavavej approved these changes Mar 27, 2020

View reviewed changes

StephanTLavavej changed the title ~~Update to Clang 10, cmake 3.17~~ Update to Clang 10, CMake 3.17 Mar 27, 2020

CaseyCarter approved these changes Mar 27, 2020

View reviewed changes

tests/std/tests/P0513R0_poisoning_the_hash/test.cpp Outdated Show resolved Hide resolved

cbezault approved these changes Mar 27, 2020

View reviewed changes

StephanTLavavej requested changes Mar 27, 2020

View reviewed changes

tests/std/expected_results.txt Show resolved Hide resolved

BillyONeal requested a review from StephanTLavavej March 28, 2020 05:25

BillyONeal mentioned this pull request Mar 28, 2020

Add testing instructions to the README #647

Merged

cbezault suggested changes Mar 28, 2020

View reviewed changes

CMakeLists.txt Outdated Show resolved Hide resolved

BillyONeal changed the title ~~Update to Clang 10, CMake 3.17~~ Update to Clang 10 Mar 28, 2020

BillyONeal changed the title ~~Update to Clang 10~~ Update to Clang 10 and Visual Studio 2019 version 16.6p2 Mar 28, 2020

BillyONeal force-pushed the update_agent branch from 366d3ab to a586c0c Compare March 28, 2020 22:51

BillyONeal requested a review from cbezault March 28, 2020 23:07

cbezault suggested changes Mar 28, 2020

View reviewed changes

README.md Outdated Show resolved Hide resolved

Update README.md

3b2bba7

cbezault approved these changes Mar 28, 2020

View reviewed changes

CaseyCarter approved these changes Mar 29, 2020

View reviewed changes

cbezault reviewed Mar 29, 2020

View reviewed changes

azure-devops/create-new-agent-image.ps1 Show resolved Hide resolved

BillyONeal merged commit ff83542 into microsoft:master Mar 29, 2020

BillyONeal deleted the update_agent branch March 29, 2020 02:15

CaseyCarter mentioned this pull request Mar 29, 2020

lerp() noexcept is now required, so we should remove comments #640

Merged

4 tasks

StephanTLavavej mentioned this pull request Aug 19, 2020

Update test configurations (remove /await and skips, add missing options) #1207

Merged

		<= (ULLONG_MAX - _Cx) / _Ax) { // unsigned long long is sufficient to store
		// intermediate calculation

Conversation

BillyONeal commented Mar 27, 2020 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

miscco commented Mar 27, 2020

Uh oh!

StephanTLavavej left a comment

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!

Choose a reason for hiding this comment

Uh oh!

BillyONeal commented Mar 27, 2020

Uh oh!

Uh oh!

CaseyCarter commented Mar 27, 2020 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

cbezault commented Mar 27, 2020

Uh oh!

CaseyCarter commented Mar 27, 2020

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

StephanTLavavej commented Mar 29, 2020

Uh oh!

StephanTLavavej commented Mar 29, 2020

Uh oh!

BillyONeal commented Mar 29, 2020

Uh oh!

StephanTLavavej commented Mar 29, 2020

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

5 participants

BillyONeal commented Mar 27, 2020 •

edited

Loading

CaseyCarter commented Mar 27, 2020 •

edited

Loading