Org-mode mailing list
 help / color / mirror / Atom feed
From: Kyle Meyer <kyle@kyleam.com>
To: Emily Bourke <undergroundquizscene@protonmail.com>
Cc: "emacs-orgmode@gnu.org" <emacs-orgmode@gnu.org>
Subject: Re: [PATCH] ox-publish.el: Speed up org-publish-cache-file-needs-publishing
Date: Mon, 04 Jan 2021 03:28:19 GMT
Message-ID: <87y2h9bctq.fsf@kyleam.com> (raw)
In-Reply-To: <OfdVIrkD5F624hYL-8RdJ5OLPKgi9CuuAwn8d6wPv2QS52VQAyvh9jSagdldoir66UVdEM1v_4oBR01jbvMC7HbvnECUEl4ShY4XplndClY=@protonmail.com>

Thank you for the patch.

Emily Bourke writes:

> I found publishing when there were no changes to be slower than
> expected. Profiling showed me that
> `org-publish-cache-file-needs-publishing' was invoking the
> `after-find-file' hooks, which I don't think is necessary.
>
> I've changed it to avoid doing that, by using `with-temp-buffer' and
> `insert-file-contents', and noticed a significant increase in speed.
>
> Is there any reason I'm missing for using `find-file-noselect' in this
> case?

Nothing jumps out to me.  For large files that are already visited, I
suppose find-file-noselect returning an existing buffer can be faster,
so relevant factors would include how many Org files a project has, how
large they are, and how many of those are visited in the current
session.  My guess is that using with-temp-buffer and
insert-file-contents would be a net gain, though that gain would be
narrowed some if the temporary buffer was put into org-mode rather than
kept in fundamental-mode (more below).

> Subject: [PATCH] ox-publish.el: Speed up
>  org-publish-cache-file-needs-publishing
>
> * lisp/ox-publish.el (org-publish-cache-file-needs-publishing): Use
> `with-temp-buffer' with `insert-file-contents' instead of
> `find-file-noselect'.  This avoids running the `after-find-file' hook,
> which can make it significantly faster.

This reads to me like after-find-file is the hook itself.  Perhaps
something like this would be clearer: "... avoids calling
after-find-file and running find-file-hook, ...".

> diff --git a/lisp/ox-publish.el b/lisp/ox-publish.el
> index 7bb2fed6e..e967286cf 100644
> --- a/lisp/ox-publish.el
> +++ b/lisp/ox-publish.el
> @@ -1290,29 +1290,26 @@ the file including them will be republished as well."
>  	 (org-inhibit-startup t)
>  	 included-files-ctime)
>      (when (equal (file-name-extension filename) "org")
> -      (let ((visiting (find-buffer-visiting filename))
> -	    (buf (find-file-noselect filename))
> -	    (case-fold-search t))
> -	(unwind-protect
> -	    (with-current-buffer buf
> -	      (goto-char (point-min))
> -	      (while (re-search-forward "^[ \t]*#\\+INCLUDE:" nil t)
> -		(let ((element (org-element-at-point)))
> -		  (when (eq 'keyword (org-element-type element))
> -		    (let* ((value (org-element-property :value element))
> -			   (filename
> -			    (and (string-match "\\`\\(\".+?\"\\|\\S-+\\)" value)
> -				 (let ((m (org-strip-quotes
> -					   (match-string 1 value))))
> -				   ;; Ignore search suffix.
> -				   (if (string-match "::.*?\\'" m)
> -				       (substring m 0 (match-beginning 0))
> -				     m)))))
> -		      (when filename
> -			(push (org-publish-cache-ctime-of-src
> -			       (expand-file-name filename))
> -			      included-files-ctime)))))))
> -	  (unless visiting (kill-buffer buf)))))
> +      (let ((case-fold-search t))
> +	(with-temp-buffer
> +	  (insert-file-contents filename)
> +	  (goto-char (point-min))

The goto-char call can be dropped now because insert-file-contents inserts
after point.

Unlike the previous code, this doesn't activate org-mode in the buffer.
That gives a speedup.  And I don't spot any code downstream that depends
on the major mode being org-mode, so it's probably safe, though perhaps
there's a subtle change in behavior here (e.g., related to syntax
table).

If org-mode isn't called, the org-inhibit-startup binding above could be
dropped.

> +	  (while (re-search-forward "^[ \t]*#\\+INCLUDE:" nil t)
> +	    (let ((element (org-element-at-point)))
> +	      (when (eq 'keyword (org-element-type element))
> +		(let* ((value (org-element-property :value element))
> +		       (filename
> +			(and (string-match "\\`\\(\".+?\"\\|\\S-+\\)" value)
> +			     (let ((m (org-strip-quotes
> +				       (match-string 1 value))))
> +			       ;; Ignore search suffix.
> +			       (if (string-match "::.*?\\'" m)
> +				   (substring m 0 (match-beginning 0))
> +				 m)))))
> +		  (when filename
> +		    (push (org-publish-cache-ctime-of-src
> +			   (expand-file-name filename))

This introduces a regression.  With the previous code, the
find-file-noselect call led to default-directory being set to the Org
file's directory, and then this expand-file call on the included file
was relative to that.  With the new code, default-directory isn't
changed, so it points to a non-existing or incorrect file unless the
current default-directory and the Org file's happen to match.

> +			  included-files-ctime)))))))))


  reply	other threads:[~2021-01-04  3:29 UTC|newest]

Thread overview: 4+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2021-01-01 19:58 Emily Bourke
2021-01-04  3:28 ` Kyle Meyer [this message]
2021-01-06 20:58   ` Emily Bourke
2021-01-07  1:11   ` Dr. Arne Babenhauserheide

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://orgmode.org

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=87y2h9bctq.fsf@kyleam.com \
    --to=kyle@kyleam.com \
    --cc=emacs-orgmode@gnu.org \
    --cc=undergroundquizscene@protonmail.com \
    /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

Org-mode mailing list

This inbox may be cloned and mirrored by anyone:

	git clone --mirror https://orgmode.org/list/0 list/git/0.git

	# If you have public-inbox 1.1+ installed, you may
	# initialize and index your mirror using the following commands:
	public-inbox-init -V2 list list/ https://orgmode.org/list \
		emacs-orgmode@gnu.org
	public-inbox-index list

Example config snippet for mirrors.
Newsgroups are available over NNTP:
	nntp://news.yhetil.org/yhetil.emacs.orgmode
	nntp://news.gmane.io/gmane.emacs.orgmode


AGPL code for this site: git clone https://public-inbox.org/public-inbox.git