http://gcc.gnu.org/bugzilla/show_bug.cgi?id=24985
--- Comment #45 from Jonathan Wakely <redi at gcc dot gnu.org> 2012-04-13 12:24:30 UTC --- (In reply to comment #42) > Yes, I like that, too. For reference, the following: > > note: candidate 'void f()' expects 0 arguments, 1 provided > void f(); void f(int,int); > ^ > note: candidate 'void f(int, int)' expects 2 arguments, 1 provided > void f(); void f(int,int); > ^ I like this for this example, but does it work as well if the function name is very long, and the "expects 2 arguments, 1 provided" is no longer in a predictable position, but pushed off to the right of a very long line? t.cc: In function 'int main()': t.cc:10:25: error: no matching function for call to 'a_long_function_name(int)' t.cc:10:25: note: candidates are: t.cc:5:6: note: void a_long_function_name() t.cc:5:6: note: candidate expects 0 arguments, 1 provided t.cc:6:6: note: void a_long_function_name(something_verbose::my_very_long_type, something_verbose::my_very_long_type) t.cc:6:6: note: candidate expects 2 arguments, 1 provided would become t.cc: In function 'int main()': t.cc:10:25: error: no matching function for call to 'a_long_function_name(int)' a_long_function_name(1); ^ t.cc:10:25: note: candidates are: t.cc:5:6: note: candidate 'void a_long_function_name()' expects 0 arguments, 1 provided void a_long_function_name(); ^ t.cc:6:6: note: candidate 'void a_long_function_name(something_verbose::my_very_long_type, something_verbose::my_very_long_type)' expects 2 arguments, 1 provided void a_long_function_name(something_verbose::my_very_long_type, something_verbose::my_very_long_type); ^ (we do already have this problem when printing ridiculous paths for stdlib headers with superfluous lib/gcc/x86_64-unknown-linux-gnu/4.8.0/../../../.. rubbish in them, is there an existing bug for that?)