[Freeciv-Dev] Re: premature GOTO abortion (PR#10719)

["Jason Short" <jdorje@xxxxxxxxxxxxxxxxxxxxx>]

<URL: http://bugs.freeciv.org/Ticket/Display.html?id=10719 >

Christian Knoke wrote:

> This is about a special problem with Goto. I consider this as a big (*sigh*)
> bug and I think, although otherwise stated, that this *can* be fixed for 2.0

Here is a patch that, fixes it.  Not very pretty though.

For both branches.

-jason

Index: server/unittools.c
===================================================================
RCS file: /home/freeciv/CVS/freeciv/server/unittools.c,v
retrieving revision 1.323
diff -u -r1.323 unittools.c
--- server/unittools.c  2 Mar 2005 19:17:51 -0000       1.323
+++ server/unittools.c  3 Mar 2005 20:13:22 -0000
@@ -3180,13 +3180,19 @@
         * next turn. */
        freelog(LOG_DEBUG, "  orders move failed (out of MP).");
        if (unit_has_orders(punit)) {
-         /* FIXME: If this was the last move, the orders will already have
-          * been freed.  This is a problem, but very hard to fix.  The
-          * way things work now the orders have to be freed before the
-          * last move is done for things like caravan popups to work (see
-          * free_unit_orders above).  So we can't wait until here to
-          * free the orders without extensive changes elsewhere. */
          punit->orders.index--;
+       } else {
+         /* FIXME: If this was the last move, the orders will already have
+          * been freed, so we have to add them back on.  This is quite a
+          * hack; one problem is that the no-orders unit has probably
+          * already had its unit info sent out to the client. */
+         punit->has_orders = TRUE;
+         punit->orders.length = 1;
+         punit->orders.index = 0;
+         punit->orders.repeat = FALSE;
+         punit->orders.vigilant = FALSE;
+         punit->orders.list = fc_malloc(sizeof(order));
+         punit->orders.list[0] = order;
        }
        send_unit_info(NULL, punit);
        return TRUE;