Re: Speed of keyboard macro execution?

[David Kastrup <dak@gnu.org>]

Eli Zaretskii <eliz@gnu.org> writes:

>> From: David Kastrup <dak@gnu.org>
>> Date: Thu, 10 Dec 2015 21:36:14 +0100
>> Cc: Eli Zaretskii <eliz@gnu.org>, emacs-devel@gnu.org
>> 
>> John Wiegley <jwiegley@gmail.com> writes:
>> 
>> >>>>>> Perry E Metzger <perry@piermont.com> writes:
>> >
>> >> On Thu, 10 Dec 2015 22:00:05 +0200 Eli Zaretskii <eliz@gnu.org> wrote:
>> >>> > From: David Kastrup <dak@gnu.org>
>> >>> > So I think it would make excellent sense to disable visual
>> >>> > positioning modes while recording and replaying keyboard macros.
>> >>> 
>> >>> No, I think you are wrong.  Think C-n and C-p again.
>> >
>> > Recording a macro has the value that it does what the user would have
>> > done, just automated. If you disable visual positioning, then in many
>> > cases it no longer does what I would have done manually.
>> 
>> Visual positioning makes no sense without visual feedback.  Can you show
>> a single example of a task to be done by a keyboard macro where visual
>> positioning would _help_ achieve that task rather than sabotage it?
>
> Even if I agree with you, it doesn't help to fix the _interactive_ use
> case of pressing C-p.  With very long lines, that can take an
> eternity.  _That_ is what we should strive to fix.

No argument here.  My proposal changing the keyboard macro execution to
switch off visual positioning was not intended as a performance fix: it
might have that effect on some benchmarks, but that's purely incidental.

Stuff like fill-paragraph will still require display engine calculations
and I'm fine with that.  Its purpose is not in any manner different
during macro recording/replay so it's definitely both useful and
consistent to consult the display engine.

With regard to visual movement, I don't see that consistency will ever
be useful in macro recording/replay.  So I'd at least like to have an
option (effective during both macro recording and replay since a
difference between the two would be both inconsistent and useless) to
get inconsistent but useful behavior here.

-- 
David Kastrup