Re: [PATCH 11/14] Remove redundant check from tui_refresh_frame_and_register_information

[Tom Tromey <tom@tromey.com>]

>>>>> "Andrew" == Andrew Burgess <aburgess@redhat.com> writes:

>> Yes, in the above, 'from_stack' is always true, so the value of
>> 'frame_info_changed_p' is immaterial.

Andrew> Except I think that's a bug in the original code.

Andrew> As your previous patch points out we already hook the register changed
Andrew> observer, so the only time that a register change will need this
Andrew> function to refresh the register display is if the frame was changed.

Andrew> If we're going to stop checking frame_info_changed_p then we might as
Andrew> well stop hooking gdb::observers::register_changed and just rely on the
Andrew> before prompt hook to update everything, right?

I don't think that will work correctly, because the TUI register window
shows changed registers in a special way.  If we rely solely on the
before-prompt hook, then any command at all will cause a redisplay that
will erase the "changed" highlighting.

This is why two phases are needed: the register window hooks into the
various things that can cause a register change, and then this is
checked before redisplay.

The patch in question here just removes some code that checks if the
frame changed.  However, if the frame did change, from_stack is already
set due to the context-changed observer.

Tom