[01:43:05] * <font color="#B0B0B0">Colourless</font> is considering doing something about that whole object disappearing problem<BR>
[01:43:41] &lt;<font color="#B0B0B0">Colourless</font>&gt; if the problem is what it seems and just use of objects by schedules after they've been deleted it should be reasonably straight forward to do something about it<BR>
[02:32:14] &lt;<font color="#0000C0">sh4rm4</font>&gt; it can't hurt if somebody's stepping through the code, but its possible the bug is gone now.<BR>
[02:32:51] &lt;<font color="#B0B0B0">Colourless</font>&gt; unlikely. A known bad bug, believed to be the cause is still there unfixed<BR>
[02:33:05] &lt;<font color="#0000C0">sh4rm4</font>&gt; which is that ?<BR>
[02:33:18] &lt;<font color="#B0B0B0">Colourless</font>&gt; use after delete/double delete<BR>
[02:34:03] &lt;<font color="#0000C0">sh4rm4</font>&gt; it should be sufficient to run it thru valgrind then<BR>
[02:34:24] &lt;<font color="#B0B0B0">Colourless</font>&gt; the circumstances to trigger it are difficult to achieve<BR>
[02:35:40] &lt;<font color="#B0B0B0">Colourless</font>&gt; but analysing how the code works shows it can be a problem. The schedules create an item and keep a pointer to it. If there is a cache out on th work the cache out code will delete the item the schedule crated. The schedule can then attempt to stuff with the deleted object because it doesn't know the object was deleted<BR>
[02:36:55] &lt;<font color="#0000C0">sh4rm4</font>&gt; so the cache out should check if the object is in the schedule list<BR>
[02:36:59] &lt;<font color="#0000C0">sh4rm4</font>&gt; and not free it<BR>
[02:37:04] &lt;<font color="#B0B0B0">Colourless</font>&gt; what it comes down to is both the schedule and the world think they own the object. <BR>
[02:37:57] &lt;<font color="#0000C0">sh4rm4</font>&gt; or cacheout does: Schedule_removeObject(x)<BR>
[02:39:24] &lt;<font color="#B0B0B0">Colourless</font>&gt; i was planning on doing something 'like' that, but a bit more complicated<BR>
[02:41:18] &lt;<font color="#B0B0B0">Colourless</font>&gt; use a notification system so users of an object are notified of it being removed/deleted from the world and form of weak reference to automate it<BR>
[04:45:18] &lt;<font color="#600060">Dominus</font>&gt; Colourless: if you take a shot at this, I promise to immediately start a playthrough of both games. especially BG as this has way more producing schedules as SI<BR>
[04:46:22] &lt;<font color="#600060">Dominus</font>&gt; sh4rm4: just believe us when we say that bug is not yet fixed :)<BR>
[12:45:10] &lt;<font color="#0000C0">sh4rm4</font>&gt; Dominus, maybe not everything is fixed, but much<BR>
[12:45:37] &lt;<font color="#0000C0">sh4rm4</font>&gt; let me reiterate: the latest bugfix fixes at least 5 different issues i reported<BR>
[12:45:58] &lt;<font color="#0000C0">sh4rm4</font>&gt; which all looked like "random failure"<BR>
[15:00:36] &lt;<font color="#600060">Dominus</font>&gt; yes, a memory leak has been fixed but it's not the root of the disappearing objects.<BR>
[15:01:10] &lt;<font color="#600060">Dominus</font>&gt; not to mention that probably over half of your problems were caused by your 'fix' ;)<BR>