unofficial mirror of emacs-devel@gnu.org 
 help / color / mirror / code / Atom feed
From: Stefan Monnier <monnier@iro.umontreal.ca>
To: emacs-devel@gnu.org
Subject: Re: Abbrev suggestions - feedback appreciated
Date: Mon, 17 Sep 2018 22:05:38 -0400	[thread overview]
Message-ID: <jwv5zz3dbny.fsf-monnier+gmane.emacs.devel@gnu.org> (raw)
In-Reply-To: CABrcCQ4goOhPUGCXwapSiy_8QyxzfuaELUpNO+NK7DR6iP9ffA@mail.gmail.com

[ Note: your patch seems to have its TABs converted into SPCs, messing
  up indentation.  I tried to fix those when quoting your code, but
  I may have mishandled some parts.  ]

> +(defcustom abbrev-suggest nil
> +  "Non-nil means we should suggest abbrevs to the user.
> +By enabling this option, if abbrev mode is enabled and if the
> +user has typed some text that exists as an abbrev, suggest to the
> +user to use the abbrev instead."
> +  :type 'boolean
> +  :group 'abbrev-mode
> +  :group 'convenience)

Does it really need to be in `convenience` as well?
What would be the disadvantages/risks of enabling it by default?

> -(defvar abbrev-expand-function #'abbrev--default-expand
> -  "Function that `expand-abbrev' uses to perform abbrev expansion.
> +(defvar abbrev-expand-function #'abbrev--try-expand-maybe-suggest
> +    "Function that `expand-abbrev' uses to perform abbrev expansion.
[...]
> +(defun abbrev--try-expand-maybe-suggest ()
> +  "Try to expand abbrev, look for suggestions if enabled.
> +This is set as `abbrev-expand-function'.  If no abbrev expansion
> +is found by `abbrev--default-expand', see if there is an abbrev
> +defined for the word before point, and suggest it to the user."
> +  (or (abbrev--default-expand)
> +      (if abbrev-suggest
> +          (abbrev-suggest-maybe-suggest)
> +        nil)))

If you put the code directly into abbrev.el, then I think you may as
well modify expand-abbrev rather than hooking into abbrev-expand-function.

The rest of the code defines abbrev--suggest-maybe-suggest rather than
abbrev-suggest-maybe-suggest.

> +(defcustom abbrev-suggest-hint-threshold 3
> +  "Threshold for when to inform the user that there is an abbrev.
> +The threshold is the number of characters that differs between
> +the length of the abbrev and the length of the expansion.  The
> +thinking is that if the expansion is only one or a few characters
> +longer than the abbrev, the benefit of informing the user is not
> +that big.  If you always want to be informed, set this value to
> +`0' or less."
> +  :type 'number
> +  :group 'abbrev-mode
> +  :group 'convenience)

Do we really need to add it to `convenience`?
Just as above, I'd recommend you just remove both :group keywords and
let the default behavior kick in (which will put it into `abbrev-mode`).

> +(defun abbrev--suggest-get-active-abbrev-expansions ()
> +  "Return a list of all the active abbrev expansions.
> +Includes expansions from parent abbrev tables."
> +  (let (expansions)
> +    (dolist (table (abbrev--suggest-get-active-tables-including-parents))
> +      (mapatoms (lambda (e)
> +                  (let ((value (symbol-value (abbrev--symbol e table))))
> +                    (when value

I think you'd be better served defining a dolist-style macro or
a mapc-style function to loop over all abbrevs without creating an
intermediate list.

> +                      (setq expansions
> +                            (cons (cons value (symbol-name e))
> +                                  expansions)))))

Aka   (push (cons value (symbol-name e)) expansions))))

> +(defun abbrev--suggest-count-words (expansion)
> +    "Return the number of words in EXPANSION.
> +Expansion is a string of one or more words."
> +    (length (split-string expansion " " t)))

Why does your code pay attention to words?

> +(defun abbrev--suggest-inform-user (expansion)
> +  "Display a message to the user about the existing abbrev.
> +EXPANSION is a cons cell where the `car' is the expansion and the
> +`cdr' is the abbrev."
> +  (run-with-idle-timer
> +   1 nil
> +   `(lambda ()
> +      (message "You can write `%s' using the abbrev `%s'."
> +               ,(car expansion) ,(cdr expansion))))

Please don't quote your lambdas.  `abbrev.el` uses lexical-binding, so
you can just write

       (run-with-idle-timer
        1 nil
        (lambda ()
          (message "You can write `%s' using the abbrev `%s'."
                   (car expansion) (cdr expansion))))

> +    (setq abbrev--suggest-saved-recommendations
> +          (cons expansion abbrev--suggest-saved-recommendations)))

Aka  (push expansion abbrev--suggest-saved-recommendations))

BTW, won't this list contain repetitions?
Maybe you should use `add-to-list` or `cl-pushnew` instead?

One more thing: I think it'd be even better to put abbrev objects (which
are implemented as symbols) in there, so you have easy accesss to
a more info than just the expansion.

> +(defun abbrev--suggest-shortest-abbrev (new current)
> +    "Return the shortest abbrev.
> +NEW and CURRENT are cons cells where the `car' is the expansion
> +and the `cdr' is the abbrev."
> +    (if (not current)
> +        new
> +      (if (< (length (cdr new))
> +             (length (cdr current)))
> +          new
> +        current)))

Maybe rather than the shortest abbrev, it would be better to choose the
abbrev with the largest difference between abbrev and expansion.

> +(defun abbrev--suggest-maybe-suggest ()
> +  "Suggest an abbrev to the user based on the word(s) before point.
> +Uses `abbrev-suggest-hint-threshold' to find out if the user should be
> +informed about the existing abbrev."
> +  (let (words abbrev-found word-count)
> +    (dolist (expansion (abbrev--suggest-get-active-abbrev-expansions))
> +      (setq word-count (abbrev--suggest-count-words (car expansion))
> +            words (abbrev--suggest-get-previous-words word-count))

Why not use something like

    (buffer-substring (- (point) (length (car expansion))) (point))?

and when it's equal to (car expansion), then maybe check that there's
a word boundary?

> +      (let ((case-fold-search t))

Some abbrevs are case-sensitive.

> +        (when (and (> word-count 0)
> +                   (string-match (car expansion) words)

(car expansion) is a string, not a regular expression, so you'd need to
regexp-quote it before passing it to string-match.

> +                   (abbrev--suggest-above-threshold expansion))
> +          (setq abbrev-found (abbrev--suggest-shortest-abbrev
> +                              expansion abbrev-found)))))
> +    (when abbrev-found
> +      (abbrev--suggest-inform-user abbrev-found))))

I suspect that abbrev--suggest-above-threshold will eliminate a fairly
large number of abbrevs for some users (e.g. those using abbrevs to fix
recurring typos) and it's a cheap test which doesn't need to allocate
any memory, so I'd recommend performing it before the calls to
abbrev--suggest-count-words and abbrev--suggest-get-previous-words.

> +(defun abbrev-suggest-try-expand-maybe-suggest ()
> +  "Try to expand abbrev, look for suggestions of no abbrev found.
> +This is the main entry to the abbrev suggest mechanism.  This is
> +set as the default value for `abbrev-expand-function'.  If no
> +abbrev expansion is found by `abbrev--default-expand', see if
> +there is an abbrev defined for the word before point, and suggest
> +it to the user."
> +  (or (abbrev--default-expand)
> +      (if abbrev-suggest
> +          (abbrev-suggest-maybe-suggest))))

Looks identical to the earlier abbrev--try-expand-maybe-suggest.


        Stefan




  reply	other threads:[~2018-09-18  2:05 UTC|newest]

Thread overview: 34+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2017-09-16  7:51 Abbrev suggestions - feedback appreciated Mathias Dahl
2017-09-16  8:46 ` Eli Zaretskii
2017-09-17 13:15   ` Mathias Dahl
2017-09-16 13:22 ` Stefan Monnier
2017-09-17 13:22   ` Mathias Dahl
2017-09-17 14:03     ` Mathias Dahl
2017-09-17 21:23     ` Stefan Monnier
2017-09-17 21:56       ` Mathias Dahl
2017-10-03 12:51         ` Kaushal Modi
2017-10-07 15:13           ` Mathias Dahl
2017-10-07 15:29             ` Stefan Monnier
2017-10-07 17:18               ` Mathias Dahl
2017-10-07 18:40                 ` Mathias Dahl
2017-10-07 22:29                   ` Ian Dunn
2017-10-07 22:44                     ` Stefan Monnier
2017-10-08 16:38                       ` Ian Dunn
2018-09-17 21:48                         ` Mathias Dahl
2018-09-18  2:05                           ` Stefan Monnier [this message]
2020-05-11 21:37                             ` Mathias Dahl
2020-05-11 22:39                               ` Mathias Dahl
2020-05-11 22:58                               ` Stefan Monnier
2020-05-16 22:10                                 ` Mathias Dahl
2020-05-16 22:22                                   ` Mathias Dahl
2020-05-17  3:13                                     ` Stefan Monnier
2020-05-17 14:59                                       ` Mathias Dahl
2020-05-17 15:45                                         ` Eli Zaretskii
2020-05-17 18:43                                           ` Mathias Dahl
2020-05-17 21:20                                             ` Stefan Monnier
2020-05-18 22:00                                               ` Mathias Dahl
2020-06-04 20:14                                                 ` Mathias Dahl
2017-10-07 22:40                 ` Stefan Monnier
2017-10-08 15:28                   ` Mathias Dahl
  -- strict thread matches above, loose matches on Subject: below --
2017-10-08  8:15 Seweryn Kokot
2017-10-08 15:25 ` Mathias Dahl

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=jwv5zz3dbny.fsf-monnier+gmane.emacs.devel@gnu.org \
    --to=monnier@iro.umontreal.ca \
    --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).