Re: [RFC] Make static tracepoint with markers more OO

[Tom Tromey <tromey at redhat dot com>]

>>>>> "Sergio" == Sergio Durigan Junior <sergiodj@redhat.com> writes:

Sergio> I have been working on this patch, and I would like some comments from
Sergio> you guys.  It basically implements new methods inside breakpoint_ops in
Sergio> order to make static tracepoint with markers (`strace -m') more OO.

I think it is very nice.  Thanks.

At first I thought that maybe the overlap between the
create_sals_from_address and decode_linespec methods could be eliminated.
But, I think the first one has to compute the text that is used
when re-evaluating, but the latter just has to re-evaluate it, and so
the savings would be minimal.

My only comment on the design is that I wonder if the handling of
pending breakpoints is done at the right place.  I assume we don't
support pending static tracepoint (not even sure if that can mean
anything) -- but I think we probably do want to support pending
SystemTap probe breakpoints.

Sergio> +  if (arg && strncmp (arg, "-m", 2) == 0 && isblank (arg[2]))

I am not sure that isblank is ok to use.  gdb doesn't use it anywhere
else.  I suggest just using the original check.

Sergio> +  /* This method will be responsible for creating a breakpoint given its SALs.
Sergio> +     Usually, it just calls `create_breakpoints_sal' (for ordinary
Sergio> +     breakpoints).  However, there may be some special cases where we might
Sergio> +     need to do some tweaks, e.g., see
Sergio> +     `strace_marker_init_or_create_breakpoint_sal'.

This function doesn't exist :)

Tom