[PATCH] respect CODESYS in `insert-file-contents-internal'.

[Nix]

On 20 Jan 2007, David Kastrup outgrape:
> For all that I know, specifying binary format image file names in an
> image descriptor is _still_ blowing up XEmacs 21.4 (and likely 21.5,
> too) once dired gets loaded.  This bug was reported at least 4 years
> ago IIRC.  Let me take a look:

Indeed. Here's the patch again, against fairly recent 21.5 CVS.

2005-02-04  Nix  <nix at esperi.org.uk>

	* fileio.c (restore_read_coding_system_unwind): Pop the coding
	system.
	(Finsert_file_contents_internal): Use the CODESYS even for files
	read by a file-name-handler.

Index: src/fileio.c
===================================================================
--- src/fileio.c.orig	2006-12-16 21:20:49.000000000 +0000
+++ src/fileio.c	2006-12-16 21:46:42.000000000 +0000
@@ -216,6 +216,15 @@
   return Fset_marker (point_marker, Qnil, Qnil);
 }
 
+/* Restore coding-system-for-read.  */
+
+static Lisp_Object
+restore_read_coding_system_unwind (Lisp_Object coding_system)
+{
+  Vcoding_system_for_read = coding_system;
+  return Qnil;
+}
+
 
 Lisp_Object Qexpand_file_name;
 Lisp_Object Qfile_truename;
@@ -3360,6 +3369,9 @@
      multiple return points make this a pain in the butt. ]] we do
      protect curbuf now. --ben */
 
+  record_unwind_protect (restore_read_coding_system_unwind, Vcoding_system_for_read);
+  Vcoding_system_for_read = codesys;
+
   codesys = get_coding_system_for_text_file (codesys, 0);
 
   if (current_buffer->base_buffer && ! NILP (visit))

-- 
`The serial comma, however, is correct and proper, and abandoning it will
surely lead to chaos, anarchy, rioting in the streets, the Terrorists
taking over, and possibly the complete collapse of Human Civilization.'