GNU bug report logs - #76985
Invert when symbol syntax class is used for ##

Previous Next

To reply to this bug, email your comments to 76985 AT debbugs.gnu.org.

Toggle the display of automated, internal messages from the tracker.

Message #5 received at submit <at> debbugs.gnu.org (full text, mbox):

From: Jonas Bernoulli <jonas <at> bernoul.li>
To: bug-gnu-emacs <at> gnu.org
Cc: Stefan Monnier <monnier <at> iro.umontreal.ca>
Subject: Invert when symbol syntax class is used for ##
Date: Wed, 12 Mar 2025 22:25:46 +0100

The `llama' package uses the `##' symbol and encourages not putting any
space in between that and the following symbol.

Since it was added three years ago, `elisp-mode-syntax-propertize'
explicitly assigns the symbol syntax class to "##", which means that
code that relies on the syntax alone (such as complete-symbol), thinks
that ##foo is one symbol, when that are actually two symbols ## and foo.

For some reason [1: 6ccc4b6bc8a] limited assigning symbol syntax for ##
to when that appears in code.  I propose we do the opposite: do NOT use
symbol syntax in code, but DO use symbol syntax in strings and comments.
Do the latter so that `##' is highlighted like other symbols would, when
quoted like that.

1: 2022-05-06 6ccc4b6bc8a14daca6b3e3250574752c90c1eb9b
   Handle elisp #-syntax better in Emacs Lisp mode

I am opening this bug-report because Stefan Monnier commented on
llama.el's advice for `elisp-mode-syntax-propertize', to ask whether
such a bug-report exists:

>> +     (syntax-propertize-rules
>> +      ;; Empty symbol.
>> +      ;; {{ Comment out to prevent the `##' from becoming part of
>> +      ;;    the following symbol when there is no space in between.
>> +      ;; ("##" (0 (unless (nth 8 (syntax-ppss))
>> +      ;;            (string-to-syntax "_"))))
>> +      ;; }}
>> +      ;; {{ As for other symbols, use `font-lock-constant-face' in
>> +      ;;    docstrings and comments.
>> +      ("##" (0 (when (nth 8 (syntax-ppss))
>> +                 (string-to-syntax "_"))))
>> +      ;; }}
>
> BTW, is there a bug-report about this?  It seems this issue is not
> really dependent on what ## is used for (except maybe for the fact that
> `llama.el` encourages the use a `##` immediately followed by another
> symbol without any intervening space).
>
>         Stefan

Coming from an override advice, the above code includes "original
vs. advice" commentary.  As a patch to Emacs, this change look like
this:

diff --git a/lisp/progmodes/elisp-mode.el b/lisp/progmodes/elisp-mode.el
@@ -269,9 +269,12 @@ elisp-mode-syntax-propertize
   (let ((case-fold-search nil))
     (funcall
      (syntax-propertize-rules
-      ;; Empty symbol.
-      ("##" (0 (unless (nth 8 (syntax-ppss))
-                 (string-to-syntax "_"))))
+      ;; Prevent ## (the empty symbol) from becoming part of a
+      ;; preceeding or following symbol by *not* explicitly assigning
+      ;; symbol syntax class here; except in comments and strings, so
+      ;; that `##' gets highlighted like other `symbols'.
+      ("##" (0 (and (nth 8 (syntax-ppss))
+                    (string-to-syntax "_"))))
       ;; Prevent the @ from becoming part of a following symbol.
       (",@" (0 (unless (nth 8 (syntax-ppss))
                  (string-to-syntax "'"))))


Jonas

Message #8 received at submit <at> debbugs.gnu.org (full text, mbox):

From: Stefan Monnier <monnier <at> iro.umontreal.ca>
To: Jonas Bernoulli <jonas <at> bernoul.li>
Cc: bug-gnu-emacs <at> gnu.org
Subject: Re: Invert when symbol syntax class is used for ##
Date: Wed, 12 Mar 2025 18:15:14 -0400

> The `llama' package uses the `##' symbol and encourages not putting any
> space in between that and the following symbol.
>
> Since it was added three years ago, `elisp-mode-syntax-propertize'
> explicitly assigns the symbol syntax class to "##", which means that
> code that relies on the syntax alone (such as complete-symbol), thinks
> that ##foo is one symbol, when that are actually two symbols ## and foo.

Do you have specific behaviors w.r.t highlighting, navigation, or
indentation, that matter specifically?

To be "technically" more correct, I suspect `##` in code should be treated
in the syntax tables as some kind of "pair of an open-paren-like thingy
and close-paren-like thingy", so that navigation with M-C-f over `##foo`
skips it in two steps.

I suspect this will lead to indentation that's different from the
one you'd prefer, e.g.

    (list (##foo
             bar)
          (##foo bar
             baz))

instead of

    (list (##foo
           bar)
          (##foo bar
                 baz))

But I think this is specific to llama's use of ## so ideally, we'd fix
that on the `llama.el` side with some `list-indent-function` magic.

> For some reason [1: 6ccc4b6bc8a] limited assigning symbol syntax for ##
> to when that appears in code.  I propose we do the opposite: do NOT use
> symbol syntax in code, but DO use symbol syntax in strings and comments.
> Do the latter so that `##' is highlighted like other symbols would, when
> quoted like that.

Sounds good to me.  I think we left ## untouched in text/comments
because we assumed that ## in comments would usually not represent ELisp
code but something else.  With the use of ## for llama, this presumption
does not seem valid any more.


        Stefan

Previous Next

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