Re: [patch] Add visible flag to breakpoints.

[Pedro Alves <pedro at codesourcery dot com>]

Below's just for the record.  I'm okay with whatever you guys
have decided.  ;-)

On Friday 08 October 2010 15:04:37, Phil Muldoon wrote:
> Pedro Alves <pedro@codesourcery.com> writes:
> 
> > On Friday 08 October 2010 13:50:34, Phil Muldoon wrote:
> >> The @var{internal} argument has no effect with watchpoints.
> >
> > Should it be an error instead (or made to work)?
> 
> I can make it an error.  I decided not to do watchpoints.  

IMO, either of these would be better API than a silent ignore.
If you go with error, the client of the API can
retry a non-internal breakpoint after e.g., printing a warning,
say.  As is, if you ever add support for internal watchpoints,
then a script has no way to tell whether such internal breakpoints
are actually supported by the gdb at hand until trying: if the flag
is just ignored, too late, the user has already seen the
breakpoint being created...

> The
> consequences of setting hidden watchpoints from a script that could turn
> out to be software watchpoints seemed a little troubling. 

IMO, it should the script writers' responsibility to weight that,
but it's also okay to not support it until someone asks for it
(though that someone would be happier if she didn't have to,
obviously :-) ).

-- 
Pedro Alves