Re: Idea for stack traces

[Bryce McKinlay <bryce at waitaki dot otago dot ac dot nz>]

Andrew Haley wrote:

>Offhand, my first reaction to this is that we'd be duplicating the
>DWARF info for no particular good effect.  Writing a tiny DWARF
>interpreter that can only access symbols and addresses which lives in
>libgcj would not be so very hard, and we could arrange for this info
>to be emitted even on targets that don't use DWARF by default.
>

The advantage of the scheme I described is that it doesn't require any 
DWARF info - only the reflective data which is available in every GCJ 
class even if there is no debugging info. Writing a small DWARF 
interpreter (we could write most of it in Java) would be the way to go 
to get more detailed info, including the source file and line number 
info for the stack trace.

regards

Bryce.