bug#19776: 25.0.50; HTML rendering is very slow

[Lars Ingebrigtsen]

Eli Zaretskii <eliz@gnu.org> writes:

> When the timer expires, it delivers a signal, but the signal handler
> only sets a flag.  The flag is checked when we call maybe_quit, which
> calls do_pending_atimers.  So a Lisp program that hogs the CPU, and
> never calls any function that calls maybe_quit, will indeed block
> atimers.  However, this is rare for real-life Lisp programs, because
> several core primitives call maybe_quit from time to time.

The use case here was for CPU-hogging Lisp code, though, like
`shr-insert-document'.  And it doesn't seem to hit any maybe_quits in
this very synthetic test:

(with-delayed-message 0.5 "Yes"
  (dotimes (i 1000000)
    (with-temp-buffer
      (shr-insert-document
       '(html nil
              (body nil
                    (table nil
                           (tr nil
                               (td nil "Here's a line ")
                               (td nil "Here's a line ")))))))))
      
Could `with-delayed-message' set a flag that would call maybe_quit more
often, somehow?

-- 
(domestic pets only, the antidote for overdose, milk.)
   bloggy blog: http://lars.ingebrigtsen.no