From: Joost Kremers <joostkremers@fastmail.fm>
To: Emacs development discussions. <emacs-devel@gnu.org>
Subject: `format` slows down my function even though it shouldn't be called at all...
Date: Sun, 15 Dec 2024 22:39:31 +0100 [thread overview]
Message-ID: <86r068wrqk.fsf@fastmail.fm> (raw)
Hi,
I have a library for parsing `.bib` files[1] and today a user reported an
issue[2] that lead me to a weird discovery. Basically, a call to `format`
that is never even executed slows down execution tremendously.
The reader part of the parser consists of a couple of functions that all
have the following structure:
```
(defun parsebib--chars (chars &optional noerror)
"Read the character at point.
CHARS is a list of characters. If the character at point matches
a character in CHARS, return it and move point, otherwise signal
an error, unless NOERROR is non-nil, in which case return nil."
(parsebib--skip-whitespace)
(if (memq (char-after) chars)
(prog1
(char-after)
(forward-char 1))
(unless noerror
(signal 'parsebib-error (list (format "Expected one of %S, got %c at position %d,%d"
chars
(following-char)
(line-number-at-pos) (current-column)))))))
```
In short, after skipping whitespace, the reader functions try to read some
element (character, keyword, etc. depending on the function), return it if
it's found and signal an error if it's not found.
The call to `signal` contains a call to `format` to provide a useful error
message. It's this `format` that slows down parsing, even though no errors
are ever signalled.
This is an excerpt from a profiler report:
============================================================
17758 81% - parsebib--chars
17755 80% - if
17749 80% - if
17737 80% - signal
17737 80% - list
17515 79% format
============================================================
This is from parsing a 28MB .bib file: The 79% of processing time spent in
`format` seems very weird to me, given that the file contains no errors and
no error is ever signalled.
After removing the `format` calls, replacing them with a simple string, the
parser runs much, much, much faster. To give an idea of the speed increase:
without `format`, the 28MB .bib file I mentioned above is parsed in 1-2
seconds (on my machine); with `format`, I don't even have enough patience
to wait for parsing to finish... (I let it run for at least 20-30 seconds
before interrupting it.)
Anyone know what's going on here? Am I missing something, or could this be
a bug in Emacs? (I'm running Emacs 29.4, BTW).
TIA
Joost
Footnotes:
[1] At https://github.com/joostkremers/parsebib
[2] https://github.com/joostkremers/parsebib/issues/34
--
Joost Kremers
Life has its moments
next reply other threads:[~2024-12-15 21:39 UTC|newest]
Thread overview: 2+ messages / expand[flat|nested] mbox.gz Atom feed top
2024-12-15 21:39 Joost Kremers [this message]
2024-12-15 22:31 ` `format` slows down my function even though it shouldn't be called at all Joost Kremers
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=86r068wrqk.fsf@fastmail.fm \
--to=joostkremers@fastmail.fm \
--cc=emacs-devel@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).