GNU bug report logs - #63861
[PATCH] pp.el: New "pretty printing" code

Previous Next

Full log

View this message in rfc822 format

From: Stefan Monnier <monnier <at> iro.umontreal.ca>
To: Thierry Volpiatto <thievol <at> posteo.net>
Cc: 63861 <at> debbugs.gnu.org
Subject: bug#63861: [PATCH] pp.el: New "pretty printing" code
Date: Tue, 20 Jun 2023 16:56:33 -0400

> So, by file, from fastest to slowest:
>
>     foo.el (0.859482743 0 0.0) (pp-buffer) t
>     foo.el (0.890402623 0 0.0) (pp-buffer) nil
>     foo.el (4.62344853 4 1.7225397670000002) (tv/pp-region (point-min) (point-max)) t
>     foo.el (4.687414465 4 1.7116580980000002) (tv/pp-region (point-min) (point-max)) nil
>     foo.el (7.932661181 1 0.3435169600000001) (pp-region (point-min) (point-max)) t
>     foo.el (196.183345212 1 0.618591124) (pp-region (point-min) (point-max)) nil
>     foo.el (2997.739238575 505 105.82851685700001) (let ((s (read (current-buffer)))) (erase-buffer) (pp-emacs-lisp-code s)) t
[...]
> We also see that `pp-emacs-lisp-code` is *much* slower.  I don't include
> other results for this function in this email because they're still
> running :-)

OK, they're done running (well, I had to re-run them because of a power
failure in between).  The tests failed on `test-load-history.el` with:

    Error: wrong-type-argument (number-or-marker-p cl--defmethod-doc-pos)

so it looks like it tickles a bug somewhere in `pp-emacs-lisp-code`.
As for the performance:

    "foo.el" (3207.572643287 505 111.459754959) (... (pp-emacs-lisp-code s)) t
    "foo.el" (121171.97145393 692 103.67438615900001) (... (pp-emacs-lisp-code s)) nil
    "test-bookmark-alist.el" (102462.563603419 5456 921.614736375) (... (pp-emacs-lisp-code s)) t
    "test-bookmark-alist.el" (191188.84323175802 7493 847.82675889) (... (pp-emacs-lisp-code s)) nil

So the `lisp-ppss` patch speeds up `pp-emacs-lisp-code` by a factor 37x
on `foo.el` and a factor a bit less than 2x for `test-bookmark-alist.el`.

We also see that `pp-emacs-lisp-code` (with the `lisp-ppss` patch) is
more than 300x slower than the new `pp-fill` code on `foo.el` and more
than 3000x slower than the new `pp-fill` code on
`test-bookmark-alist.el`.

Admittedly, these are not cases for which that code was designed (these
files hold data rather than code).


        Stefan

Previous Next

GNU bug tracking system
Copyright (C) 1999 Darren O. Benham, 1997,2003 nCipher Corporation Ltd, 1994-97 Ian Jackson.