Zugg Software :: View topic - [2.03] Automapper crash with memLookup exeption

Enchanter Joined: 09 Sep 2007 Posts: 605

I spam with this bug report for 2 days, it seems that you had not receive any of that reports.

I use the automapper in Fast-Follow mode, walk around with keyboard, keeping an eye on map. At least 1 time a day the automapper crashes with the following message:
memLookup: Dataset is not in edit mode.

I noticed that it crashes in 3 groups of adjaсent rooms: two groups in one zone and one group in another. I see nothing special in these rooms, all exits marked as tested... In these rooms I slow down my movement speed for about 1 room per second, it somehow help to pass them without crash (but not always). My normal movement speed is about 6-10 rooms per second.

From bugreport.txt:

exception class : EDatabaseError
exception message : memLookup: Dataset is not in edit mode.

Looks like the CMUD tries to update some fields in DBset, but it (set) is locked for write.

call stack is always the same:

Posted: Wed Sep 26, 2007 9:43 pm

The crash dumps that you sent are probably in my database, but are mixed in with hundreds of others. And since you mentioned it in the other thread, I thought it might be useful to get more information on it.

Are there any Room Scripts set for the rooms that cause the problem? Does entering the rooms cause any of your other triggers to fire? Looks like you pressed the keypad key while a trigger was in the middle of executing. Any idea what trigger it might have been in the middle of?

Enchanter Joined: 09 Sep 2007 Posts: 605

I do not have any room triggers. I'd try to add some simple room script, but it ends up with another bug report to you (unable delete room script completely, mapper auto add "All=" string, which cause error message), if you remember. In any way, rooms with "All=" and rooms which cause crash are not the same, they are completely different.

At least, 1 trigger executes with every room passed - prompt trigger, plus any other trigger from my information triggers class may fire on the route.

MUD not always reacts fast, sometimes I may send 3-5 keypresses before it start spit to me with room descriptions - it all depends on server load and cant be predicted.

The last thing - I may send you my map database, it is not big yet.

Posted: Thu Sep 27, 2007 4:44 pm

You might also want to try testing it Offline. Use the Offline action to open your session and map. Then set your position on the map to one of the "problem" rooms. Then try to quickly move back and forth using the Keypad macros. The room descriptions will display to the MUD window, but you won't get any MUD prompt, so your prompt triggers won't fire. If it still crashes when doing this Offline, then definitely send me your map because it should be easy to reproduce and fix. If it only happens when connected to the MUD, then it will be harder to figure out what is causing it.

Enchanter Joined: 09 Sep 2007 Posts: 605

I've uncheck option "Mark non tested exits as grey" - it works good only for links to another Z level or if link lead to not drawn room. If I have 2 rooms connected at only 1->2 way was tested, the link draws as tested, though in spreadsheet view link 2->1 still indicate not tested. So I must correct myself: there are NotTested links, but I always receive a crash while move along tested links.

I walk around while offline, even run. Without any crashes.

Heh, we definitely not looking for easy ways. Twisted Evil

Wizard Joined: 14 Aug 2004 Posts: 1269

Try turning off triggers globally and see if you still get crashes. If you don't get crashes with that, try turning them on except for your prompt trigger and try that.

Wizard Joined: 14 Aug 2004 Posts: 1269

Arde, did you try my suggestion?

Enchanter Joined: 09 Sep 2007 Posts: 605

Yeah, but I spent not so much time... Not a crash for 10 minutes. Although, I've got one later, with all triggers on. As I wrote, it happens 1 or 2 times per several hours and it is not so easy to catch it. That all was in 2.03, I have not see this crash yet in 2.04, but I not use it so much as 2.03. More, I had "All=" bug in 2.03 that Zugg fixed in 2.04; each "All=" error message produced +1 stopped thread, may be this somehow increase risk to get a crash.