FYI, I have all of the CorrectTypo call sites updated and everything working (including fixit mode, which didn't require mucking with the CXXScopeSpec [and said mucking screws up diagnostic messages]); I'm just waiting on guidance for resolving the issues with basic.lookup.argdep/p4.cpp. I'm tempted to just fix up the expectations in basic.lookup.argdep/p4.cpp, except the old error message is far more useful than the new and misleading/bogus messages and I'd like to figure out a way to keep the old behavior if possible without severely crippling the missing-namespace corrections.
Cheers, Kaelyn On Thu, Jun 23, 2011 at 5:11 PM, Kaelyn Uhrain <[email protected]> wrote: > + // FIXME: Don't do the lookups if argument dependent lookup is >> required as >> + // this will break >> test/CXX/basic/basic.lookup/basic.lookup.argdep/p4.cpp >> >> Is this FIXME still relevant, given that we're doing the appropriate >> requiresADL check? >> > > I've found a problem with the simple ADL check in CorrectTypo... with it, > common cases we want to suggest corrections for such as: > > namespace fizbin { > namespace nested { bool moreFoobar() { return true; } } > } > void foo() { > moreFoobar(); > } > > give: > > tmp.cpp:5:3: error: use of undeclared identifier 'moreFoobar' > moreFoobar(); > ^ > 1 error generated. > > instead of: > > tmp.cpp:5:3: error: use of undeclared identifier 'moreFoobar'; did you mean > 'fizbin::nested::moreFoobar'? > moreFoobar(); > ^~~~~~~~~~ > fizbin::nested::moreFoobar > tmp.cpp:2:27: note: 'fizbin::nested::moreFoobar' declared here > namespace nested { bool moreFoobar() { return true; } } > > but without the check, basic.lookup.argdep/p4.cpp reports: > > error: 'error' diagnostics seen but not expected: > Line 35: no viable conversion from 'B::B' to 'C::C' > > error: 'note' diagnostics seen but not expected: > Line 23: 'C::func' declared here > Line 22: candidate constructor (the implicit copy constructor) not > viable: no known conversion from 'B::B' to 'const C::C &' for 1st argument > Line 23: passing argument to parameter here > 4 errors generated. > > Ideas? > > - Kaelyn >
_______________________________________________ cfe-commits mailing list [email protected] http://lists.cs.uiuc.edu/mailman/listinfo/cfe-commits
