On Wed, Jul 18, 2012 at 3:35 PM, Aaron Ballman <[email protected]>wrote:
> On Wed, Jul 18, 2012 at 5:07 PM, Richard Smith <[email protected]> > wrote: > > On Wed, Jul 18, 2012 at 2:56 PM, Aaron Ballman <[email protected]> > > wrote: > >> > >> Author: aaronballman > >> Date: Wed Jul 18 16:56:43 2012 > >> New Revision: 160455 > >> > >> URL: http://llvm.org/viewvc/llvm-project?rev=160455&view=rev > >> Log: > >> Fixing up a test case that was failing due to the lack of -std=c++11 > > > > > > What was the failure? This testcase doesn't seem to have anything to do > with > > C++11, and had been unchanged for ~2 years. > > I'm still not certain why this is different, but this test case used > to pass on Windows when run from vanilla cmd.exe, and now it's stopped > (though it does run fine from a VS 2010 command prompt). The problem > is a string of errors dealing with auto: > > Command 0 Stderr: > error: 'error' diagnostics seen but not expected: > Line 414: 'auto' not allowed in function return type > Line 414: expected ';' at end of declaration list > Line 414: 'auto' not allowed in function return type > Line 414: expected ';' at end of declaration list > Line 414: expected parameter declarator > Line 414: expected ')' > Line 414: '_Fn' cannot be the name of a parameter > Line 414: expected ';' at end of declaration list > Line 197: use of undeclared identifier 'static_assert'; did you mean > 'static_c > ast'? > Line 214: use of undeclared identifier 'static_assert'; did you mean > 'static_c > ast'? > Line 238: expected parameter declarator > Line 238: expected ')' > Line 442: function definition does not declare parameters > Line 488: no viable conversion from 'int' to 'std::exception_ptr' > Line 527: unknown type name 'nullptr_t' > Line 532: unknown type name 'nullptr_t' > error: 'note' diagnostics seen but not expected: > Line 414: to match this '(' > Line 237: to match this '(' > Line 450: candidate constructor not viable: no known conversion from > 'int' to > 'const std::exception_ptr &' for 1st argument > 19 errors generated. > > So I do agree that the -std=c++11 change isn't a good fix (I'll back > it out), but I'm stumped as to why this has suddenly stopped working > when it was previously fine. This test indirectly includes <typeinfo>. Presumably your system now has a <typeinfo> which requires C++11. I'm surprised you don't also see failures in the 8 tests in test/CodeGenCXX which include <typeinfo>.
_______________________________________________ cfe-commits mailing list [email protected] http://lists.cs.uiuc.edu/mailman/listinfo/cfe-commits
