Re: utils.c: Sign-extend addresses if required by the target

[Daniel Jacobowitz <drow@false.org>]

On Mon, Dec 17, 2007 at 05:37:44PM +0000, Maciej W. Rozycki wrote:
> On Sun, 16 Dec 2007, Daniel Jacobowitz wrote:
> 
> > OK.  This routine is used by MI, but should not be - a long term TODO
> > item is to change the way we present frame IDs to MI frontends to not
> > be based on the stack pointer, and then its use will go away.  But
> > that's not going to happen too soon.
> 
>  What about Insight?  Not that I would die for it, but the function is 
> called from a few places in gdbtk/generic/gdbtk-{cmds,stack,varobj}.c and 
> given the amount of development done on Insight these days I would imagine 
> these uses are not going to go away anytime soon.

Right.  I'm not going to touch it; I don't know what Insight uses it
for, or whether it should.  But it's probably using it for the same
reason MI is, since Insight predates varobjs.

I try not to break Insight knowingly, but if we go to clean up MI
frame identification and that breaks Insight, I'll leave it to the
Insight developers to clean up.  That's the downside of being so
tightly tied into GDB.

-- 
Daniel Jacobowitz
CodeSourcery