Fix test suite issues discovered by Clang 15 #3135
Merged
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
I validated the STL against Clang 15.0.1, which may be shipping Soon(TM) with Visual Studio.
* Clang's "initialized but unused variable" warning got smarter, and noticed that `do_random_test::gen_data` in `tests/tr1/include/tspec_random.h` initialized `int n = 0`, incremented `n` some number of times, and never used the resulting value. I've removed the offending variable.
* Clang diagnosed the subsumption test in `tests/std/P0898R3_concepts` as ill-formed due to ambiguity. The test does something like:
```c++
constexpr bool f(auto&&) { return false; }
constexpr bool f(integral auto) { return true; }
static_assert(f(42));
```
intending that `f(integral auto)` is called with the belief that its contraints subsume those of `f(auto&&)`. This was derived from similar subsumption tests I wrote for concepts in cmcstl2 and range-v3 which GCC's TS-era concepts implementation once accepted, but current GCC and Clang both reject it. I suspect MSVC and Clang 14 were wrong to accept the code and the call _should_ be ambiguous, but I need to dig into the wording for partial ordering of function templates before filing a bug against MSVC. In any case, we can change `f(auto&&)` to `f(auto)` without impacting the test - the arguments used in the test all have scalar types - so let's do so.
* Clang diagnosed hundreds of unqualified calls to `move` and `forward` in our tests thanks to a new `unqualified-std-cast-call` warning. I don't address these in this PR, I've instead opened microsoft#3134 for us to discuss precisely how we do want to address it.
strega-nil-ms
approved these changes
Sep 29, 2022
StephanTLavavej
approved these changes
Oct 11, 2022
|
I'm mirroring this to the MSVC-internal repo - please notify me if any further changes are pushed. |
|
Thanks for scouting ahead and fixing these failures! 😻 🛠️ ✨ |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
I validated the STL against Clang 15.0.1, which may be shipping Soon(TM) with Visual Studio.
Clang's "initialized but unused variable" warning got smarter, and noticed that
do_random_test::gen_dataintests/tr1/include/tspec_random.hinitializedint n = 0, incrementednsome number of times, and never used the resulting value. I've removed the offending variable.Clang diagnosed the subsumption test in
tests/std/P0898R3_conceptsas ill-formed due to ambiguity. The test does something like:c++ constexpr bool f(auto&&) { return false; } constexpr bool f(integral auto) { return true; } static_assert(f(42));intending thatf(integral auto)is called with the belief that its contraints subsume those off(auto&&). This was derived from similar subsumption tests I wrote for concepts in cmcstl2 and range-v3 which GCC's TS-era concepts implementation once accepted, but current GCC and Clang both reject it. I suspect MSVC and Clang 14 were wrong to accept the code and the call should be ambiguous, but I need to dig into the wording for partial ordering of function templates before filing a bug against MSVC. In any case, we can changef(auto&&)tof(auto)without impacting the test - the arguments used in the test all have scalar types - so let's do so.Clang diagnosed hundreds of unqualified calls to
moveandforwardin our tests thanks to a newunqualified-std-cast-callwarning. I don't address these in this PR, I've instead opened test: hundreds of failures with Clang 15.0.1 #3134 for us to discuss precisely how we do want to address it.