Look for cased character classes when deciding on case-fold-search, #'isearch lisp/ChangeLog addition: 2015-03-11 Aidan Kehoe <kehoea@parhasard.net> * isearch-mode.el: * isearch-mode.el (isearch-fix-case): Use the new #'no-case-regexp-p function if treating ISEARCH-STRING as a regular expression; otherwise, use the [[:upper:]] character class. * isearch-mode.el (isearch-no-upper-case-p): Removed. * isearch-mode.el (with-caps-disable-folding): Removed. These two haven't been used since 1998. * occur.el (occur-1): Use #'no-case-regexp-p here. * replace.el (perform-replace): Don't use #'no-upper-case-p, use #'no-case-regexp-p or (string-match "[[:upper:]]" ...) as appropriate. * simple.el: * simple.el (no-upper-case-p): Removed. This did two different things, and its secondary function (examining regular expressions) just became much more complicated; move the regular expression functionality to its own function, use character classes when examining non-regular-expressions instead. The code to look for character classes, and the design decision that this should be done, are from GNU, thank you Stefan Monnier. * simple.el (no-case-regexp-p): New. Given a REGEXP, return non-nil if it has nothing to suggest an interactive user wants a case-sensitive search. * simple.el (with-search-caps-disable-folding): * simple.el (with-interactive-search-caps-disable-folding): Update both these macros to use #'no-case-regexp-p.

--- a/lisp/replace.el	Wed Mar 11 15:06:05 2015 +0000
+++ b/lisp/replace.el	Wed Mar 11 18:06:15 2015 +0000
@@ -563,7 +563,11 @@
 	 ;; XEmacs addition
 	 (qr-case-fold-search
 	  (if (and case-fold-search search-caps-disable-folding)
-	      (no-upper-case-p search-string regexp-flag)
+              (if regexp-flag
+                  (no-case-regexp-p search-string)
+                (save-match-data
+                  (let (case-fold-search)
+                    (not (string-match "[[:upper:]]" search-string)))))
 	    case-fold-search))
 	 (message
 	  (if query-flag