Warning c6319 in Visual Studio

[Cvelth]

Description

Visual Studio marks asserts with warning c6319.

Steps to reproduce

• $ git clone https://github.com/Cvelth/cpp_starter_fork
• $ cmake.exe -Htest -Bbuild -G "Visual Studio 16 2019"
• open build/GreeterTests.sln
• navigate to GreeterTests/Source Files/greeter.cpp
  
  

Extra information

• doctest version: v2.3.7
• Operating System: Windows 10 (1909, build 18363.778)
• Compiler+version: Microsoft Visual Studio 2019 (16.5.1) with MSVC 19.25.28611.0

[onqtam]

Seems that there's no easy way to make all the static code analysis tools happy...

I used the do { ... } while((void)0, 0) trick to avoid warnings because do { ... } while(false) was a problem with some compilers a few years back (conditional expression is constant)... But it seems that the latest MSVC complains about this. I'll try again with do { ... } while(false).

[Cvelth]

Can confirm that do { ... } while(false) resolves the issue and works great. Not sure about other compilers, though.

[onqtam]

I'll probably do a push with such a change in a test branch to see what the CI reports (if all supported compilers are fine with that) but I can't promise that will happen anytime soon.

[Cvelth]

I doesn't seem to cause any problems (proof).
The only failing build seems to be a compiler issue (a typo in a header file), which should be fixed with the release of visual studio 16.6 (msvc 14.26), as it's already fixed in 16.6 preview 2. Proofs: 1, 2.

EDIT: <concepts> header is implicitly included by <utility>.
EDIT_2: Exact issue: (link)

[onqtam]

Well ok, that sounds reasonable. You can open a PR from your branch and we'll see if all the builds pass (including travis & appveyor - besides just github actions)

[Cvelth]

Travis CI and AppVeyor both pass.

I was hesitant to open PR as it does fail on a single build-system.

[onqtam]

I don't see how your changes could be triggering that failure - I also just looked at the previous github actions builds of doctest and seems that this failure is already present: https://github.com/onqtam/doctest/actions

So a PR would be just fine - just make sure to open it against the dev branch - seems that you made the changes on top of master.

[Cvelth]

Should I PR changes after the target assemble_single_header is run, or just doctest_fwd.h changes?

[onqtam]

preferably after that target has run - I like to keep the resulting header in sync, but it's not a deal-breaker.