From: Michael Heerdegen <michael_heerdegen@web.de>
To: Eli Zaretskii <eliz@gnu.org>
Cc: eggert@cs.ucla.edu, 37321@debbugs.gnu.org
Subject: bug#37321: 27.0.50; Excessive gc in a use case (el-search)
Date: Wed, 09 Oct 2019 16:47:58 +0200 [thread overview]
Message-ID: <87o8yqynz5.fsf@web.de> (raw)
In-Reply-To: <83a7abxubl.fsf@gnu.org> (Eli Zaretskii's message of "Tue, 08 Oct 2019 16:03:58 +0300")
Eli Zaretskii <eliz@gnu.org> writes:
> > in a way that gc doesn't lower the speed that much
>
> AFAIK, the latter can only be done by changing your algorithms to
> produce less garbage.
I tried to find out what code produces the most garbage. It turned out
that ca. 50% of garbage was generated by code that prevents infinite
recursion when recursing into nested structures. I use hash tables to
collect visited objects, and the hash tables cause the garbage. I tried
to reuse hash tables and clear them after each use, but this makes the
code much slower than what I win from gc.
But 99,9% of el-searched code isn't cyclic, so the effort is for nothing
most of the time. Is there an efficient way to find out if a given
object is cyclic?
For now I try with this:
(lambda ()
(save-excursion)
(goto-char (point-min))
(search-forward-regexp "#[0-9+]=[^\"]" nil t))
(all treated objects are read from a buffer, so I can inspect the
contents) and get good results but it feels a bit hackish.
Thanks,
Michael.
next prev parent reply other threads:[~2019-10-09 14:47 UTC|newest]
Thread overview: 38+ messages / expand[flat|nested] mbox.gz Atom feed top
2019-09-06 13:52 bug#37321: 27.0.50; Excessive gc in a use case (el-search) Michael Heerdegen
2019-09-07 14:23 ` Michael Heerdegen
2019-09-07 15:30 ` Michael Heerdegen
2019-09-08 1:11 ` Paul Eggert
2019-09-08 14:52 ` Michael Heerdegen
2019-09-08 15:25 ` Michael Heerdegen
2019-09-14 8:04 ` Paul Eggert
2019-09-14 8:37 ` Eli Zaretskii
2019-09-14 8:52 ` Paul Eggert
2019-09-14 9:57 ` Eli Zaretskii
2019-09-14 17:57 ` Paul Eggert
2019-09-14 18:16 ` Eli Zaretskii
2019-09-15 4:33 ` Richard Stallman
2019-09-16 23:53 ` Michael Heerdegen
2019-09-17 0:55 ` Paul Eggert
2019-09-21 0:41 ` Michael Heerdegen
2019-09-21 0:46 ` Michael Heerdegen
2019-09-21 6:19 ` Paul Eggert
2019-09-17 12:47 ` Noam Postavsky
2019-09-21 0:44 ` Michael Heerdegen
2019-09-25 9:42 ` Michael Heerdegen
2019-09-25 20:37 ` Paul Eggert
2019-09-26 11:42 ` Michael Heerdegen
2019-09-26 12:14 ` Eli Zaretskii
2019-09-26 13:03 ` Michael Heerdegen
2019-10-08 8:43 ` Michael Heerdegen
2019-10-08 9:09 ` Eli Zaretskii
2019-10-08 9:11 ` Michael Heerdegen
2019-10-08 9:19 ` Eli Zaretskii
2019-10-08 11:12 ` Michael Heerdegen
2019-10-08 12:11 ` Eli Zaretskii
2019-10-08 12:38 ` Michael Heerdegen
2019-10-08 13:03 ` Eli Zaretskii
2019-10-09 14:47 ` Michael Heerdegen [this message]
2019-10-09 15:33 ` Eli Zaretskii
2019-10-09 20:53 ` Paul Eggert
2019-10-10 10:58 ` Michael Heerdegen
2019-10-08 9:22 ` Michael Heerdegen
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
List information: https://www.gnu.org/software/emacs/
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=87o8yqynz5.fsf@web.de \
--to=michael_heerdegen@web.de \
--cc=37321@debbugs.gnu.org \
--cc=eggert@cs.ucla.edu \
--cc=eliz@gnu.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
Code repositories for project(s) associated with this public inbox
https://git.savannah.gnu.org/cgit/emacs.git
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).