Re: The cost of stack traces

[Casey Marshall <csm at gnu dot org>]

On May 11, 2006, at 7:07 AM, Andrew Haley wrote:

> One thing that seems almost to have got completely lost in all this
> talk about how to make stack traces faster is the sheer pointlessness
> of it all.  Many (if not most) of the installed gcj-compiled libraries
> don't have debuginfo and usually no-one cares about the line numbers
> anyway.
>
> One thing we could do is detect libraries with no debuginfo and from
> that point onwards not bother looking for it.
>
> By all means let's make the traces faster but there are surely better
> uses of our time.

I don't agree that it's pointless. Lots of people (probably for the  
worse, but that isn't the point) rely on source and line numbers in  
stack traces to diagnose problems. It really is the best compromise  
to not having __LINE__ and __FILE__ in Java.

In the end, this is a feature that developers using GCJ (who are your  
users!) are going to want. Because you personally find it useless  
isn't what they (I!) are going to want to hear.

I also personally think it's cool for GCJ-compiled programs to  
reflect on themselves -- even down to the level of looking through  
shared libraries. That's why having a tool internal to the library  
for doing that is neat.

Thanks.

Attachment: PGP.sig
Description: This is a digitally signed message part