Add #'query-coding-region tests; do the work necessary to get them running. lisp/ChangeLog addition: 2008-12-28 Aidan Kehoe <kehoea@parhasard.net> * coding.el (default-query-coding-region): Declare using defun*, so we can #'return-from to it on encountering a safe-charsets value of t. Comment out a few debug messages. (query-coding-region): Correct the docstring, it deals with a region, not a string. (unencodable-char-position): Correct the implementation for non-nil COUNT, special-case a zero value for count, treat it as one. Don't rely on dynamic scope when calling the main lambda. * unicode.el (unicode-query-coding-region): Comment out some debug messages here. * mule/mule-coding.el (8-bit-fixed-query-coding-region): Comment out some debug messages here. * code-init.el (raw-text): Add a safe-charsets property to this coding system. * mule/korean.el (iso-2022-int-1): * mule/korean.el (euc-kr): * mule/korean.el (iso-2022-kr): Add safe-charsets properties for these coding systems. * mule/japanese.el (iso-2022-jp): * mule/japanese.el (jis7): * mule/japanese.el (jis8): * mule/japanese.el (shift-jis): * mule/japanese.el (iso-2022-jp-1978-irv): * mule/japanese.el (euc-jp): Add safe-charsets properties for all these coding systems. * mule/iso-with-esc.el: Add safe-charsets properties to all the coding systems in here. Comment on the downside of a safe-charsets value of t for iso-latin-1-with-esc. * mule/hebrew.el (ctext-hebrew): Add a safe-charsets property for this coding system. * mule/devanagari.el (in-is13194-devanagari): Add a safe-charsets property for this coding system. * mule/chinese.el (cn-gb-2312): * mule/chinese.el (hz-gb-2312): * mule/chinese.el (big5): Add safe-charsets properties for these coding systems. * mule/latin.el (iso-8859-14): Add an implementation for this, using #'make-8-bit-coding-system. * mule/mule-coding.el (ctext): * mule/mule-coding.el (iso-2022-8bit-ss2): * mule/mule-coding.el (iso-2022-7bit-ss2): * mule/mule-coding.el (iso-2022-jp-2): * mule/mule-coding.el (iso-2022-7bit): * mule/mule-coding.el (iso-2022-8): * mule/mule-coding.el (escape-quoted): * mule/mule-coding.el (iso-2022-lock): Add safe-charsets properties for all these coding systems. src/ChangeLog addition: 2008-12-28 Aidan Kehoe <kehoea@parhasard.net> * file-coding.c (Fmake_coding_system): Document our use of the safe-chars and safe-charsets properties, and the differences compared to GNU. (make_coding_system_1): Don't drop the safe-chars and safe-charsets properties. (Fcoding_system_property): Return the safe-chars and safe-charsets properties when asked for them. * file-coding.h (CODING_SYSTEM_SAFE_CHARSETS): * coding-system-slots.h: Make the safe-chars and safe-charsets slots available in these headers. tests/ChangeLog addition: 2008-12-28 Aidan Kehoe <kehoea@parhasard.net> * automated/query-coding-tests.el: New file, testing the functionality of #'query-coding-region and #'query-coding-string.

--- a/lisp/mule/iso-with-esc.el	Sat Aug 09 13:15:09 2008 +0200
+++ b/lisp/mule/iso-with-esc.el	Sun Dec 28 14:46:24 2008 +0000
@@ -28,6 +28,10 @@
 
 ;;; Code:
 
+;; It is not particularly reasonable that iso-latin-1-with-esc has a
+;; value of t for the safe-charsets property. We discourage its use,
+;; though, and this behaviour is compatible with GNU.
+
 ;;;###autoload
 (define-coding-system-alias 'iso-latin-1-with-esc 'iso-2022-8)
 
@@ -38,6 +42,7 @@
    charset-g1 latin-iso8859-2
    charset-g2 t
    charset-g3 t
+   safe-charsets (ascii latin-iso8859-2)
    mnemonic "MIME/Ltn-2"))
 
 ;;;###autoload
@@ -47,6 +52,7 @@
    charset-g1 latin-iso8859-3
    charset-g2 t
    charset-g3 t
+   safe-charsets (ascii latin-iso8859-3)
    mnemonic "MIME/Ltn-3"))
 
 ;;;###autoload
@@ -56,6 +62,7 @@
    charset-g1 latin-iso8859-4
    charset-g2 t
    charset-g3 t
+   safe-charsets (ascii latin-iso8859-4)
    mnemonic "MIME/Ltn-4"))
 
 ;;;###autoload
@@ -63,6 +70,7 @@
  'iso-latin-9-with-esc 'iso2022
   "ISO 4873 conforming 8-bit code (ASCII + Latin 9; aka Latin-1 with Euro)"
   '(mnemonic "MIME/Ltn-9"		; bletch
+    safe-charsets (ascii latin-iso8859-15)
     eol-type nil
     charset-g0 ascii
     charset-g1 latin-iso8859-15
@@ -76,6 +84,7 @@
    charset-g1 latin-iso8859-9
    charset-g2 t
    charset-g3 t
+   safe-charsets (ascii latin-iso8859-9)
    mnemonic "MIME/Ltn-5"))
 
 ;;;###autoload
@@ -86,6 +95,7 @@
    charset-g1 cyrillic-iso8859-5
    charset-g2 t
    charset-g3 t
+   safe-charsets (ascii cyrillic-iso8859-5)
    mnemonic "ISO8/Cyr"))
 
 ;;;###autoload
@@ -97,6 +107,7 @@
    charset-g2 t
   charset-g3 t
    no-iso6429 t
+   safe-charsets (ascii hebrew-iso8859-8)
    mnemonic "MIME/Hbrw"))
 
 ;;;###autoload
@@ -106,6 +117,7 @@
    charset-g1 greek-iso8859-7
    charset-g2 t
    charset-g3 t
+   safe-charsets (ascii greek-iso8859-7)
    mnemonic "Grk"))
 
 ;; ISO 8859-6 is such a useless character set that it seems a waste of
@@ -201,5 +213,6 @@
    charset-g2 t
    charset-g3 t
    no-iso6429 t
+   safe-charsets (ascii arabic-iso8859-6)
    mnemonic "MIME/Arbc"))