Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!watmath!clyde!caip!lll-crg!nike!ucbcad!ucbvax!CC5.BBN.COM!jr
From: jr@CC5.BBN.COM (John Robinson)
Newsgroups: net.emacs
Subject: Re: print-buffer args error? (GNU)
Message-ID: <8608011352.AA21078@ucbvax.Berkeley.EDU>
Date: Fri, 1-Aug-86 09:46:02 EDT
Article-I.D.: ucbvax.8608011352.AA21078
Posted: Fri Aug  1 09:46:02 1986
Date-Received: Sat, 2-Aug-86 09:18:24 EDT
References: <624@gcc-milo.ARPA>
Sender: daemon@ucbvax.BERKELEY.EDU
Organization: The ARPA Internet
Lines: 36

It looks as though the function print-region-1 will screw up when it
has to call expand to diddle the tabs, which it does if the tab stops
are other than every 8 characters.  It fails to recompute the region
boundaries (formals start and end) after filtering the print stuff
through expand.  In fact, it looks as if the normal (non-error) case
will chop off (fail to print) the end of the buffer or region.

Here's an attempt at a fixed print-region-1.  Since we don't run lpr
here, I can't really test it.  It might be better to construct a shell
pipeline with expand and lpr to avoid the extra process step (and the
bug!); that is an exercise for the reader or maintainer.

/jr

(defun print-region-1 (start end switches)
  (save-excursion
   (message "Spooling...")
   (if (/= tab-width 8)
       (let ((oldbuf (current-buffer)))
	(set-buffer (get-buffer-create " *spool temp*"))
	(widen) (erase-buffer)
	(insert-buffer-substring oldbuf)
	(set-mark (max start end))
	(call-process-region start end "expand"
			     t t nil
			     (format "-%d" tab-width))
	(if (> start end)
	    (setq start (mark))
	  (setq end (mark)))))
   (apply 'call-process-region
	  (nconc (list start end "lpr"
		       nil nil nil
		       "-J" (concat (buffer-name) " Emacs buffer")
		       "-T" (concat (buffer-name) " Emacs buffer"))
		 switches))
   (message "Spooling...done")))