Re: RFC: stack trace generation

[Mark Wielaard <mark at klomp dot org>]

Hi,

On Sun, 2004-07-18 at 08:17, Casey Marshall wrote:
> Also, what is the status of Mark's interpreted frame info patch?

Ehm, yeah. You refer to this patch I assume:
http://gcc.gnu.org/ml/java-patches/2004-q2/msg00365.html (against 3.4)
http://gcc.gnu.org/ml/java-patches/2004-q2/msg00411.html (against head)

I haven't found/made time to do what Tom and Bryce asked me to do yet.
They suggested measuring the impact of either making pc into a local or
only setting frame_desc.pc when a instruction can actually throw. And
some simple cleanup of the patch (see thread). Someday I will have time
for that...

Also Anthony Green claims that the patch doesn't work for him when
playing with Eclipse 3.0. I haven't replicated his experiment. With
Eclipse 2.1 there was no problem however (and also no regressions or new
Mauve failures). So the patch can probably also use some real world
testing. Again, one day...

If someone has time for any of the above, please do! The chances of me
finding time any of the above within the next two/three weeks are slim.

Cheers,

Mark

Attachment: signature.asc
Description: This is a digitally signed message part