On Fri, 7 Apr 2000, matthew green wrote:

>    But what about the C++ problem reported in PR 8465?
> 
> 
> i think this best fix for this is to undefine dwarf2 unwind info,
> which has the same effect as -fsjlj-exceptions.
> 
> 
> + /* We don't support this (yet?); use the old way */
> + #undef DWARF2_UNWIND_INFO
> + #define DWARF2_UNWIND_INFO 0

I don't think so (although it is the fastest fix...) sjlj exceptions is
much worse preformance wise, and AFIK there will be a ABI change if we
change the exception mechanism later, so we want to DTRT now.

I'll take a look at it this weekend...

   /Krister