[PATCH] xemacs-21.5: Proposed fix for `call-process-region' bug uncovered tracking down suspected crypt++.el bug

[Adrian Aichner]

Greetings!

crypt++.el works fine under non-Mule XEmacs 21.4.15 with a minor
modification I made (proposed patch under discussion with crypt++.el
maintainer).

In 21.5-b16 Mule XEmacs, visisting a gpg encrypted file leads an empty
buffer!

I have tracked this down to code-process.el, which I believe is
currently broken because it does not consider insertion of process
output at an arbitrary point in buffer.

It also got the MARKER-TYPE argument for the s and e markers mixed up,
if I understand this correctly.

Here is the proposed patch which I would like to submit to your
scrutiny and better yet, your testing.

Works fine during minimal testing.

Please advise,

Adrian

xemacs-21.5 ChangeLog patch:
Diff command:   cvs -q diff -U 0
Files affected: lisp/ChangeLog

Index: lisp/ChangeLog
===================================================================
RCS file: /pack/xemacscvs/XEmacs/xemacs/lisp/ChangeLog,v
retrieving revision 1.554
diff -u -U0 -r1.554 ChangeLog
--- lisp/ChangeLog	8 Mar 2004 15:22:44 -0000	1.554
+++ lisp/ChangeLog	23 Mar 2004 00:23:15 -0000
@@ -0,0 +1,6 @@
+2004-03-23  Adrian Aichner  <adrian at xemacs.org>
+
+	* code-process.el (call-process-region): Correct start and end
+	marker construction, delete regions before and after insertion
+	point.
+

xemacs-21.5 source patch:
Diff command:   cvs -f -z3 -q diff -u -N
Files affected: lisp/code-process.el

Index: lisp/code-process.el
===================================================================
RCS file: /pack/xemacscvs/XEmacs/xemacs/lisp/code-process.el,v
retrieving revision 1.12
diff -u -r1.12 code-process.el
--- lisp/code-process.el	10 Nov 2002 16:07:13 -0000	1.12
+++ lisp/code-process.el	23 Mar 2004 00:05:18 -0000
@@ -169,16 +169,22 @@
   ;; a temporary file was used to pass the text to call-process.  Now that
   ;; we don't do that, we delete the text afterward; if it's being inserted
   ;; in the same buffer, make sure we track the insertion, and don't get
-  ;; any of it in the deleted region if insertion happens at either end
-  ;; of the region.
-  (let ((s (and deletep (copy-marker start t)))
-	(e (and deletep (copy-marker end))))
+  ;; any of it in the deleted region.  We keep marker s before the
+  ;; insertion and e afterward.  Finally we delete the regions before
+  ;; and after the insertion.
+  (let ((s (and deletep (copy-marker (point))))
+	(e (and deletep (copy-marker (point) t))))
     (let ((retval
 	   (apply #'call-process program (list (current-buffer) start end)
 		  buffer displayp args)))
-      ;; If start and end were the same originally, s will be beyond e now
-      (if (and deletep (> e s))
-	  (delete-region s e))
+      ;; If start and end were the same originally, e will be beyond s now
+      (when (and deletep (> e s))
+	;; APA: Is it always correct to honor narrowing, which affects
+	;; (point-min) and (point-max)?
+	;; Delete region before insertion.
+	(delete-region (point-min) s)
+	;; Delete region after insertion.
+	(delete-region e (point-max)))
       retval)))
 
 (defun start-process (name buffer program &rest program-args)

-- 
Adrian Aichner
 mailto:adrian at xemacs.org
 http://www.xemacs.org/