Re: RFA: Breakpoint infrastructure cleanups [0/8]

[Jim Ingham <jingham@apple.com>]

Yeah, but if I told gdb to set breakpoints at *0x....., and sometimes 
those breakpoints would just stick at that address, and other times 
they would move around, that would be disconcerting to me.  If the 
breakpoint is not going to be at a fixed address, we oughtn't to refer 
to it as if it were.

Jim

On Oct 16, 2003, at 11:59 PM, Eli Zaretskii wrote:

>> From: Jim Ingham <jingham@apple.com>
>> Date: Thu, 16 Oct 2003 12:03:11 -0700
>>
>> I would be careful to stay away from turning "logically" specified
>> breakpoints (by which I mean specified on function name or source
>> location) into addresses to the user.  Even between rerunnings of the
>> same executable a library's load address can shift, causing the 
>> address
>> to move.  gdb can probably still make the equivalency between the
>> breakpoints - most slides are rigid, for instance.  But the address
>> doesn't show this.
>
> I didn't say that GDB should _store_ the address that the user types
> in order to disambiguate the place where to put the trap.  It is just
> a means to tell GDB which of the possibilities to take.  It has an
> advantage of being natural to GDB users, since you can put a
> breakpoint on a specific address in current versions of GDB.
>
> After the trap was put, if GDB can solve the problem of moving
> addresses (as it does that now), it can also solve the problem we are
> discussing here.
>
--
Jim Ingham                                   jingham@apple.com
Developer Tools
Apple Computer