Closed gamesh411 closed 5 years ago
Unfortunately, seems like we have a strange clang-tblgen assertion error in CI. :(
I could build, and clang-test succeeded. OSX specific tests could still fail, as I cannot test on OSX locally. Maybe the CI should incorporate a test on OSX as well.
I could build, and clang-test succeeded. OSX specific tests could still fail, as I cannot test on OSX locally. Maybe the CI should incorporate a test on OSX as well.
Yes, I agree.
But this time the CI error is with table-gen: http://ci1.codechecker.gic.ericsson.se:8080/job/ctu_pipeline_pullrequest/152/console
run tests
Are these ODR warnings reported only from AST import related code or in other case when code is compiled? Is it possible that a case that was a compile error before the change is now only warning but the generated code is not correct? (The objective-c test cases may be the problematic ones.)
Are these ODR warnings reported only from AST import related code or in other case when code is compiled?
This is valid. The only use case of StructuralEquivalenceContext out of the ASTImporter is in Sema::hasStructuralCompatLayout which is called from Sema::ActOnDuplicateDefinition which is used from ParseEnumSpecifier and from ParseClassSpecifier.
Now, making this a warning affects Clang. But there is not any lit test which we break here. This is not a fatal error, so the original behavior can be achieved by setting -WError=odr.
create upstream patch
By default all odr-related diagnostics are considered warnings belonging to Diagnostic Group "Odr". They can be propagated to error level with -Werror=odr.