bug#38170: 27.0.50; [PATCH] rx.el: Fix misplaced "char" identifiers.

bug#38170: 27.0.50; [PATCH] rx.el: Fix misplaced "char" identifiers.

[Zhu Zihao]

[-- Attachment #1: Type: text/plain, Size: 179 bytes --]


New rx implementation introduce some bug:

1. (rx char) throw an error "Unknown rx symbol char"
2. (rx (char ?A ?B)) return a regexp match A or B.

This patch aims to fix this.


[-- Attachment #2: 0001-rx.el-Fix-misplaced-char-identifiers.patch --]
[-- Type: text/plain, Size: 1385 bytes --]

From 1b5d28b8b2bcbf91ad32991d3602de45ef9bef76 Mon Sep 17 00:00:00 2001
From: Zhu Zihao <all_but_last@163.com>
To: bug-gnu-emacs@gnu.org
Date: Sun, 10 Nov 2019 20:02:44 +0800
Subject: 27.0.50; [PATCH] rx.el: Fix misplaced "char" identifiers.

---
 lisp/emacs-lisp/rx.el | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/lisp/emacs-lisp/rx.el b/lisp/emacs-lisp/rx.el
index 52a35ffa2a7..c5f341a7b0a 100644
--- a/lisp/emacs-lisp/rx.el
+++ b/lisp/emacs-lisp/rx.el
@@ -153,7 +153,7 @@ rx--translate-symbol
   (pcase sym
     ;; Use `list' instead of a quoted list to wrap the strings here,
     ;; since the return value may be mutated.
-    ((or 'nonl 'not-newline 'any) (cons (list ".") t))
+    ((or 'nonl 'not-newline 'any 'char) (cons (list ".") t))
     ((or 'anychar 'anything)      (cons (list "[^z-a]") t))
     ('unmatchable                 (rx--empty))
     ((or 'bol 'line-start)        (cons (list "^") 'lseq))
@@ -857,7 +857,7 @@ rx--translate-form
     (pcase (car form)
       ((or 'seq : 'and 'sequence) (rx--translate-seq body))
       ((or 'or '|)              (rx--translate-or body))
-      ((or 'any 'in 'char)      (rx--translate-any nil body))
+      ((or 'any 'in)            (rx--translate-any nil body))
       ('not-char                (rx--translate-any t body))
       ('not                     (rx--translate-not nil body))
 
-- 
2.24.0

bug#38170: 27.0.50; [PATCH] rx.el: Fix misplaced "char" identifiers.

[Mattias Engdegård]

Thanks for the report! However, there may be a misunderstanding:

(char ...) as an alias for (any ...) was a documented feature of the old implementation and was therefore preserved in the re-write; it cannot be removed without a very good justification.

'char' as an alias for 'not-newline' was never documented in the first place and because of the rather intuitive meaning, it was considered to be an implementation accident and therefore not included.

Therefore I'd rather not apply the proposed changes.

bug#38170: 27.0.50; [PATCH] rx.el: Fix misplaced "char" identifiers.

[Zhu Zihao]

On Mon, 11 Nov 2019 21:37:09 +0800,
Mattias Engdegård wrote:
> 
> Thanks for the report! However, there may be a misunderstanding:
> 
> (char ...) as an alias for (any ...) was a documented feature of the old implementation and was therefore preserved in the re-write; it cannot be removed without a very good justification.
> 
> 'char' as an alias for 'not-newline' was never documented in the first place and because of the rather intuitive meaning, it was considered to be an implementation accident and therefore not included.
> 
> Therefore I'd rather not apply the proposed changes.

Thanks for your explanation. I' m going to adjust my code instead :)