Re: [RFA 1/4] Explicit locations

[Tom Tromey <tromey at redhat dot com>]

>>>>> "Keith" == Keith Seitz <keiths@redhat.com> writes:

Keith> This is a reboot of a series of patches I submitted several months
Keith> ago. The feature I am attempting to introduce will essentially allow
Keith> users to bypass the linespec parser.

I'm sorry for the long delay on this.

Keith> +/* An enumeration of the various ways to specify a stop event
Keith> +   location (used with create_breakpoint).  */
Keith> +
Keith> +enum event_location_type
Keith> +  {

There's some inconsistency in Emacs and GNU here, but I think we
generally try to put that open brace in column 0.

I think event_location_type and friends could go in new .h and .c files.
breakpoint.c is already one of the largest files we have.

 
Keith>  int
Keith>  create_breakpoint (struct gdbarch *gdbarch,
Keith> -		   char *arg, char *cond_string,
Keith> +		   struct event_location *location, char *cond_string,
Keith>  		   int thread, char *extra_string,

I think the intro comment needs an update.
The ownership of the passed-in location wasn't totally clear to me.

E.g., there's at least one spot initializing a location with a
non-malloc'd string; which I guess is fine provided the ownership rules
are clearly spelled out.


I think it would be good to spell out the plan for -probe arguments.
It seems like it would be good to integrate them.  For one thing, maybe
that would make it so dprintf could work with probe locations...

Tom