From: Mattias Engdegård <mattiase@acm.org>
Date: Sun, 12 May 2024 09:22:23 +0000 (+0200)
Subject: Safer and faster string sorting
X-Git-Url: http://git.eshelyaron.com/gitweb/?a=commitdiff_plain;h=45e04484357737ef225a4cff8ee6a8de68cb36c6;p=emacs.git

Safer and faster string sorting

* lisp/emacs-lisp/cl-seq.el (cl-sort):
Don't use string mutation when sorting the characters in a string.
This avoids O(n^2) run time and makes it future-safe.

(cherry picked from commit 38091e43bee991920db81263fe56da444a123fae)
---

diff --git a/lisp/emacs-lisp/cl-seq.el b/lisp/emacs-lisp/cl-seq.el
index 42f54603899..60d72d8657d 100644
--- a/lisp/emacs-lisp/cl-seq.el
+++ b/lisp/emacs-lisp/cl-seq.el
@@ -668,7 +668,10 @@ This is a destructive function; it reuses the storage of SEQ if possible.
 \nKeywords supported:  :key
 \n(fn SEQ PREDICATE [KEYWORD VALUE]...)"
   (if (nlistp cl-seq)
-      (cl-replace cl-seq (apply 'cl-sort (append cl-seq nil) cl-pred cl-keys))
+      (if (stringp cl-seq)
+          (concat (apply #'cl-sort (vconcat cl-seq) cl-pred cl-keys))
+        (cl-replace cl-seq
+                    (apply #'cl-sort (append cl-seq nil) cl-pred cl-keys)))
     (cl--parsing-keywords (:key) ()
       (if (memq cl-key '(nil identity))
 	  (sort cl-seq cl-pred)