- A warning is now sent to the oper if (s)he tries to add a (G)ZLINE on *@host.

(G)ZLINES should have an ipmask, not a hostmask, because they are processed BEFORE
  any dns lookups are done.

Changes

@@ -716,3 +716,9 @@
         So use with care on mixed-version nets.
 - Fixed /(G)ZLINE [nick] placing the *line on *@host instead of *@IP, reported by
   Snake (#0002246).
+- A warning is now sent to the oper if (s)he tries to add a (G)ZLINE on *@host.
+  (G)ZLINES should have an ipmask, not a hostmask, because they are processed BEFORE
+  any dns lookups are done. Therefore any (g)zlines placed will probably work
+  (but not necessarily) for like an hour (or whatever TTL), but after that the
+  (ab)user can get in again so this is usually not what you want ;).
+  I suppose I'll add a FAQ entry about this.

src/modules/m_tkl.c

@@ -447,6 +447,21 @@ DLLFUNC int  m_tkl_line(aClient *cptr, aClient *sptr, int parc, char *parv[], ch
 			hostmask = usermask;
 			usermask = "*";
 		}
+		
+		if ((*type == 'z') || (*type == 'Z'))
+		{
+			/* It's a (G)ZLINE, make sure the user isn't specyfing a HOST.
+			 * Just a warning for now, but perhaps in 3.2.4 we should make this an error.
+			 */
+			for (p=hostmask; *p; p++)
+				if (isalpha(*p))
+				{
+					sendnotice(sptr, "WARNING: (g)zlines should be placed on user@IPMASK, not user@hostmask "
+					                 "(this is because (g)zlines are processed BEFORE a dns lookup is done)");
+					break;
+				}
+		}
+		/* set 'p' right for later... */
 		p = hostmask-1;
 	}
 	else