From bug-octave-request at bevo dot che dot wisc dot edu Mon Nov 25 00:23:25 2002
Subject: Function cache too tenacious
From: "John W. Eaton" <jwe at bevo dot che dot wisc dot edu>
To: Julius Smith <jos at w3k dot org>
Cc: bug-octave at bevo dot che dot wisc dot edu
Date: Mon, 25 Nov 2002 00:22:54 -0600

On 22-Nov-2002, Julius Smith <jos at w3k dot org> wrote:

| I'm running Octave version 2.1.36 on a Linux PC, and I encountered a what 
| appears to be a bug in function caching behavior. Briefly described, I 
| often debug functions by commenting out their declaration line at the top 
| and invoking them by name only (after setting up arguments using 
| assignments).  This helps because all the function local variables become 
| available in the top-level scope where they can be inspected.  The problem 
| I found is that Octave seems to be trying to use the function definition 
| after I have "overridden it" with a script version of the same name.
| Even more surprising, "clear functions" does not fix the problem.

I don't think 2.1.36 had "clear functions" as a special case, so all
it would try to do is clear a symbol called "functions".

| However, an 
| explicit "clear function-name" does work, so that's a workaround.  All of 
| this is illustrated in the diary output below:

The following patch (relative to the current CVS) seems to fix the
problem.

jwe


src/ChangeLog:

2002-11-25  John W. Eaton  <jwe at bevo dot che dot wisc dot edu>

	* pt-stmt.cc (tree_statement::eval): Allow the lookup to execute
	script files.  If script file has been executed, don't bother to
	call expr->rvalue ().


Index: src/pt-stmt.cc
===================================================================
RCS file: /usr/local/cvsroot/octave/src/pt-stmt.cc,v
retrieving revision 1.17
diff -u -r1.17 pt-stmt.cc
--- src/pt-stmt.cc	20 Nov 2002 16:56:49 -0000	1.17
+++ src/pt-stmt.cc	25 Nov 2002 06:18:54 -0000
 at @ -111,25 +111,32 @@
 	  // or not the expression will take care of binding ans and
 	  // printing the result.
 
+	  // XXX FIXME XXX -- it seems that we should just have to
+	  // call expr->rvalue () and that should take care of
+	  // everything, binding ans as necessary?
+
 	  bool do_bind_ans = false;
 
+	  bool script_file_executed = false;
+
 	  if (expr->is_identifier ())
 	    {
-	      bool script_file_executed = false;
-
 	      tree_identifier *id = static_cast<tree_identifier *> (expr);
 
-	      id->do_lookup (script_file_executed, false);
+	      id->do_lookup (script_file_executed, true);
 
 	      do_bind_ans = id->is_function ();
 	    }
 	  else
 	    do_bind_ans = (! expr->is_assignment_expression ());
 
-	  retval = expr->rvalue (nargout);
+	  if (! script_file_executed)
+	    {
+	      retval = expr->rvalue (nargout);
 
-	  if (do_bind_ans && ! (error_state || retval.empty ()))
-	    bind_ans (retval(0), pf);
+	      if (do_bind_ans && ! (error_state || retval.empty ()))
+		bind_ans (retval(0), pf);
+	    }
 	}
 
       unwind_protect::run ();



-------------------------------------------------------------
Octave is freely available under the terms of the GNU GPL.

Octave's home on the web:  http://www.octave.org
How to fund new projects:  http://www.octave.org/funding.html
Subscription information:  http://www.octave.org/archive.html
-------------------------------------------------------------