View Issue Details
| ID | Project | Category | Date Submitted | Last Update | |
|---|---|---|---|---|---|
| 0001941 | AI War 1 / Classic | Crash/Exception | Dec 8, 2010 9:37 pm | Dec 9, 2010 11:24 am | |
| Reporter | Draco18s | Assigned To | keith.lamothe | ||
| Status | resolved | Resolution | fixed | ||
| Product Version | 4.046 | ||||
| Summary | 0001941: Piles of Unhandled Errors | ||||
| Description | Non-crashing, they just periodically went "what's up?" and wrote a ton to the error log. | ||||
| Tags | No tags attached. | ||||
| Internal Weight | |||||
| duplicate of | 0001874 | resolved | keith.lamothe | UnhandledErrors during gameplay |
| duplicate of | 0001890 | resolved | keith.lamothe | Unhandled Exception errors reported |
|
|
|
|
|
|
|
|
Looks a lot like that other issue. -----------------------------------Error----------------------------------- Log String: FixedUpdate: System.NullReferenceException: Object reference not set to an instance of an object at ForegroundUnitType.GetAttackPowerAgainstThis (.ForegroundObject AttackingUnit, Int32 ArmorBoost, Int32 ArmorDamage, Int32 Health) [0x00000] in <filename unknown>:0 at ForegroundObject.GetAttackPowerAgainstThis (.ForegroundObject AttackingObject) [0x00000] in <filename unknown>:0 at OtherObject.DoShotHit () [0x00000] in <filename unknown>:0 at Game.RunOneCycleOfSimulation () [0x00000] in <filename unknown>:0 at Game.RunNextCycle (Boolean DoRendering) [0x00000] in <filename unknown>:0 at Game.RunFrame () [0x00000] in <filename unknown>:0 at MainCameraLogic.FixedUpdate () [0x00000] in <filename unknown>:0 Stack Trace: at ForegroundUnitType.GetAttackPowerAgainstThis (.ForegroundObject AttackingUnit, Int32 ArmorBoost, Int32 ArmorDamage, Int32 Health) [0x00000] in <filename unknown>:0 at ForegroundObject.GetAttackPowerAgainstThis (.ForegroundObject AttackingObject) [0x00000] in <filename unknown>:0 at OtherObject.DoShotHit () [0x00000] in <filename unknown>:0 at Game.RunOneCycleOfSimulation () [0x00000] in <filename unknown>:0 at Game.RunNextCycle (Boolean DoRendering) [0x00000] in <filename unknown>:0 at Game.RunFrame () [0x00000] in <filename unknown>:0 at MainCameraLogic.FixedUpdate () [0x00000] in <filename unknown>:0 |
|
|
Ah yes, the [i]one[/i] I didn't look for first. :P |
|
|
Yep, already mounted on the trophy wall; hopefully 4.047 will be out today. Not that it will contain any fewer crashes, what with changing the central structures storing things like [i]all foreground objects[/i], but we have to keep things on their toes ;) |
|
|
Yeah, it will be out this morning. I decided not to release it last night because there were still a few more things I needed to check, and I didn't want to have a possibly-buggy new version out while I was asleep (not that I'm seeing any crashes at this stage, haven't seen many for a while, but the effects might be more subtle). On the plus side, those changes have really made a dent in our ongoing goal for a mixture of speed and RAM usage lowering, so it was certainly worthwhile as a detour yesterday. I hadn't planned on it, but I was implementing one certain feature that had to check a dictionary a few hundred times a second, and so I decided to swap that over, and it kind of snowballed from there. ;) |
|
|
Oh yea, anything touching that global dictionary isn't a snowball, it's an avalanche ;) But great to have that transition done; Dictionary<K,T> is great for general purpose work but it's not well suited to critical-path performance on CPU, perm-alloc, or transient-alloc. |
|
|
To be fair, on .NET it seemed to be pretty fine. It's just on Mono that it results in transient allocations to this degree. Actually, I didn't change the global dictionary, btw -- that can't really be done, since the global set of ship ids could well go into the hundreds of millions. I changed away DOZENS of dictionaries on Player, AIPlayer, Enums, DatabaseContent, and a few other locations. Those were the ones that were actually being hit way more frequently, as the global ships dictionary is only used in rarer lookup cases. Actually, come to that... I might be able to replace that with just a list... will look at. |
|
|
Yea, the global set can get highly sparse due to neinzul ships, blade spawners, etc. Not to mention normal play. A List<T> can be heap-static, but I don't see how it would get around the problem of just using an array: it's sparse. Unless you're planning to have the lookup be linear-time, which I would advise against ;) Though honestly I don't know how the dictionary lookup is done. I'd just go with a custom datastructure, but you may not want that kind of time going into that right now. |
|
|
For now, tabling that. I was thinking of a hybrid thing like I use on the AI Loop, but it turns out that that dictionary is only enumerated exceedingly rarely, and not during normal gameplay cycles. So it's mostly a moot point as we're just using it as an actual dictionary and even that is fairly uncommon (commands to ships aside). |
| Date Modified | Username | Field | Change |
|---|---|---|---|
| Dec 8, 2010 9:37 pm | Draco18s | New Issue | |
| Dec 8, 2010 9:37 pm | Draco18s | File Added: Screenshot_2010_12_07_22_33_41.png | |
| Dec 8, 2010 9:37 pm | Draco18s | File Added: UnhandledErrors.txt | |
| Dec 8, 2010 9:38 pm | Toll | Note Added: 0005735 | |
| Dec 8, 2010 9:38 pm | Toll | Relationship added | duplicate of 0001874 |
| Dec 8, 2010 9:39 pm | Toll | Relationship added | duplicate of 0001890 |
| Dec 8, 2010 9:40 pm | Draco18s | Note Added: 0005736 | |
| Dec 9, 2010 8:53 am | keith.lamothe | Note Added: 0005756 | |
| Dec 9, 2010 8:53 am | keith.lamothe | Status | new => resolved |
| Dec 9, 2010 8:53 am | keith.lamothe | Resolution | open => fixed |
| Dec 9, 2010 8:53 am | keith.lamothe | Assigned To | => keith.lamothe |
| Dec 9, 2010 8:53 am | keith.lamothe | Note Edited: 0005756 | |
| Dec 9, 2010 10:59 am | Chris_McElligottPark | Note Added: 0005769 | |
| Dec 9, 2010 11:09 am | keith.lamothe | Note Added: 0005770 | |
| Dec 9, 2010 11:11 am | Chris_McElligottPark | Note Added: 0005771 | |
| Dec 9, 2010 11:19 am | keith.lamothe | Note Added: 0005775 | |
| Dec 9, 2010 11:24 am | Chris_McElligottPark | Note Added: 0005777 | |
| Apr 14, 2014 9:27 am | Chris_McElligottPark | Category | Bug - Crash or Exception => Crash/Exception |
