a bug in rtext_key()?

[Krzysztof Czaja]

Hi,

I am more of a casual user then Pd expert, and what
follows could be just plainly wrong...

The symptoms: while editing text in object/message
boxes I often experienced (what seems to be) a buggy
behaviour -- the apparent bug is usually only annoying,
but it is also capable of forcing Pd to crash.

The remedy (attached): a small patch over rtext_key()
of g_rtext.c.

Now the question: am I supposed to report such findings
through pd-list, directly to the author, or anywhere else?

Best to you all,

  Krzysztof Czaja
-------------- next part --------------
--- g_rtext.c	Mon Oct  2 00:38:38 2000
+++ ../patched/g_rtext.c	Sun Apr  8 18:00:13 2001
@@ -390,8 +390,10 @@
 	ndel = x->x_selend - x->x_selstart;
 	for (i = x->x_selend; i < x->x_bufsize; i++)
     	    x->x_buf[i- ndel] = x->x_buf[i];
-	newsize = x->x_bufsize - ndel;
+	newsize = x->x_bufsize - ndel; // because of the bug below newsize may be -1
+	// int newsize passed to size_t formal argument...
 	x->x_buf = resizebytes(x->x_buf, x->x_bufsize, newsize);
+	// ... and after passing negative newsize we get x->x_buf nulled
 	x->x_bufsize = newsize;
 
 	if (n == '\n' || isprint(n))
@@ -407,16 +409,16 @@
 	x->x_selend = x->x_selstart;
 	x->x_glist->gl_editor->e_textdirty = 1;
     }
-    else if (!strcmp(keysym->s_name, "Right"))
+    else if (!strcmp(keysym->s_name, "Right")) // ->
     {
-    	if (x->x_selend == x->x_selstart)
+    	if (x->x_selend == x->x_selstart && x->x_selend < x->x_bufsize)
 	    x->x_selend = x->x_selstart = x->x_selstart + 1;
 	else
 	    x->x_selstart = x->x_selend;
     }
-    else if (!strcmp(keysym->s_name, "Left"))
+    else if (!strcmp(keysym->s_name, "Left")) // <-
     {
-    	if (x->x_selend == x->x_selstart && x->x_selend < x->x_bufsize - 1)
+    	if (x->x_selend == x->x_selstart && x->x_selend > 0)
 	    x->x_selend = x->x_selstart = x->x_selstart - 1;
 	else
 	    x->x_selend = x->x_selstart;