http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55881
--- Comment #5 from Manuel López-Ibáñez <manu at gcc dot gnu.org> 2013-01-08 11:14:49 UTC --- (In reply to comment #4) > (In reply to comment #2) > > > Well - confirmed. Unlikely to be fixed. > > That's _very_ unfortunate. It makes the pragma almost useless in practice. The pragma can only work if it somehow knows that location 5:19 is "expanded" (inlined) from the location of "return i.foo(n)" since the code checks that the location included in the warning is within the range of the pragma and "5:19" is clearly not. This could be implement in the same way as we currently handle macro expansions, however, it won't be a trivial amount of work, and it is quite unlikely that any current developer has the free time and the interest necessary to do it themselves. If you really want this feature, you have to either try to implement it yourself or convince someone to do it for you. Or you may do "n = n" and silence the warning.
