Re: Performance degradation from long lines

[Alexander Shukaev]

On 10/26/18 6:18 PM, Stefan Monnier wrote:
> >            (let ((inhibit-field-text-motion t))
> >              (catch 'break
> >                (while (not (eobp))
> >                  (when (> (- (line-end-position 1) (point))
> >                           global-longlines-threshold)
> >                    (throw 'break t))
> >                  (forward-line 1)))))))))
>
> I think you can speed this up with the code below (I'd expect a speed up
> of about a factor 2 since (line-end-position 1) does about the same as
> (forward-line 1)):
>
>      (catch 'break
>        (while
>            (unless (eobp)
>              (or (> (+ (point) global-longlines-threshold)
>                        (progn (forward-line 1) (point)))
>                  (throw 'break t)))))

Thanks for the tip!

> > (defun inhibit-set-auto-mode-line-number-display-limit-function ()
> >    (and (integerp line-number-display-limit)
> >         (> (buffer-size) line-number-display-limit)))
> > (add-hook 'inhibit-set-auto-mode-functions
> >            #'inhibit-set-auto-mode-line-number-display-limit-function)
>
> Hmm... there seems to be a discrepancy between the identifier and the
> code: it seems unrelated to line-number(s).  Maybe some left-over from
> a previous version of the code?
>
>
>          Stefan

You are right, sorry about `line-number-display-limit', it should of 
course be something more reasonable, e.g. (8 * 1024 * 1024) for 8GiG cap.