Re: insert-file-contents and format-decode

[martin rudalics <rudalics@gmx.at>]

 >     !   (let (value size old-undo)
 >     !     ;; Record only one undo entry for the insertion.  Inhibit point-motion and
 >     !     ;; modification hooks as with `insert-file-contents'.
 >     !     (let ((inhibit-point-motion-hooks t)
 >     ! 	  (inhibit-modification-hooks t))
 >     !       ;; Don't bind `buffer-undo-list' to t here to assert that
 >     !       ;; `insert-file-contents' may record whether the buffer was unmodified
 >     !       ;; before.
 >     !       (let ((format-alist nil))
 >     ! 	(setq value (insert-file-contents filename nil beg end))
 >     ! 	(setq size (nth 1 value)))
 >     !       (when (consp buffer-undo-list)
 >     ! 	(let ((head (car buffer-undo-list)))
 >     ! 	  (when (and (consp head)
 >     ! 		     (equal (car head) (point))
 >     ! 		     (equal (cdr head) (+ (point) size)))
 >     ! 	    ;; Remove first entry from `buffer-undo-list', we shall insert
 >     ! 	    ;; another one below.
 >     ! 	    (setq old-undo (cdr buffer-undo-list)))))
 >
 > That pateh is basically good, but it could be simpler.  Instead of
 > letting insert-file-contents produce an undo entry and taking it off
 > and taking it apart, it would be better to inhibit undo around
 > insert-file-contents and generate the undo entry from scratch in all
 > cases.  When no format decoding is done, VALUE has the size you need.

It's complicated because I may have to create the (t HIGH . LOW) entry
to indicate that the buffer previously had "unmodified" status.  With my
solution I let `insert-file-contents' take care of that automatically.
With your solution I would have to create it manually and care about the
modification time myself.  Is it worth the trouble?