https://gcc.gnu.org/bugzilla/show_bug.cgi?id=58123
Aldy Hernandez <aldyh at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
Status|NEW |RESOLVED
Resolution|--- |FIXED
--- Comment #12 from Aldy Hernandez <aldyh at gcc dot gnu.org> ---
Fix committed to trunk. With this patch, we have a consistent debugging
experience, as was originally expected/reported in the PR.
I agree, that eventually we should stop jumping back to the definition of the
class instance. Perhaps that can be filed as an enhancement request. If that
is the case, then what we want to revisit is whether the destructor call should
be marked with is_stmt=false or whether we should avoid marking locationless
FINALLY statements with the location of the TRY.
