Bug Reports MAY2010a

[Thothie]

Eewww... Yeah, going straight to pack bypasses game_deploy, doesn't it? Grrr... Didn't think of that.

Meh, I'll have to call the function again from game_putinpack, or somethin. I guess that explains why this still happens occasionally.

 

[Keldorn]

Things also vanish if you transfer them inside containers without taking them out.

Had issues where I gave new players a set of rusty weapons inside a backpack only to have the items vanish on map change.

 

[Thothie]

*sigh* Yeah, transferring packs is straight out, lest you take every item in the pack and put it in your hands one at a time, in addition to the pack itself.

Yeah, for now, protip: make sure you put each item to be transfered to you in your hands, before you put it in a pack.

 

[Thothie]

Finally got around to looking into this Lodagond-4 horror thing... Bloody thing is just a standard horror with a different submodel, more HP, and the same poison spore cloud used by a half dozen other beasties... Doesn't use any sprites that aren't already precache'd by the player, so, and doesn't any crash when I spawn him by himself... Thought maybe it was the spore cloud sprites hitting the water, but I can't duplicate the crash on Lodagond-4 on my own either... Hrmm... Dunno wtf is going on with this. :/

It's just these horrors yes? Seems I did a few Calruin runs without crashing from those horrors. :\

 

[J-M v2.5.5]

I managed to duplicate the crash alone recently (holy shit) and I'm pretty sure what's causing it:
Remember back in the days, when Cleicert was just out, that it'd lag incredibly because all the lightning skeletons wanted to cast their lightning beam at the same time?
I think the problem in lodagond-4 is: all the horrors see you at the same time and then they all cast their poison clouds simultaneously. I'm pretty sure the map is just overflowing from all the poison clouds that suddenly pop up. You should try having them cast the poison cloud at a random moment, and not necessarily when they first spot a player.

 

[Thothie]

Well, with the Lightning Skeles, that made sense, because they were creating new server side beams and damaging you every 0.1 seconds (litteraly)... But the horrors here... That's four (max) client side scripts and one damage strike per second (discounting bites, since the poison effect does not stack). Hrmm... Didn't crash during my three solo runs, but they were all on a listen server.

But they don't crash until they spit the clouds at ye, eh? That gives me something at least...

 

[J-M v2.5.5]

Let me put it this way: I can't remember a single crash that did not involve the poison clouds.

 

[Keldorn]

Let me put it this way: I can't remember a single crash that did not involve the poison clouds.

This is pretty much how it has gone for me as well.

Lol noxious spores.

 

[J-M v2.5.5]

More like obnoxious AMIRITE

Edit: While you're looking at Lodagond... Whenever Maldora dies:
1) He drops a tomahawk
2) He talks to you
3) A chest shows up
4) The artifact chest might show up
5) Some flag on your char is changed to avoid lodagond-4 farming

Because all of this happens at the same time, it's not uncommon for lodagond-4 to crash the instant Maldora dies. This is even worse than crashing in the "blue magic room". Perhaps you could set it up so that Maldora drops his tomahawk one second after being beaten, the regular chest shows up two seconds after he has been beaten, the artifact chest shows up three seconds after he's down, and so on...?

 

[Thothie]

RE: Horror crashies

Hrmm... Okies, I did the delay thing, re-wrote the npc spore cloud script (which meant re-writing the npc acid and npc poison cloud scripts too), and fixed a bug I found in the poison spore DOT... About an hour spent on stabs in the dark. :\

Only other thing I could think it maybe, is that the regular spit projectile the horror, and several other monsters, use bursts into a not-often-used sprite on impact via a server-side effect script. I'll have to dig deeper, but if that sprite isn't normally precached, it may cause a crash on maps near the sprite limit (which Lodagond-4 may or may not be.)

RE: Maldora crashies

Shall examine. Suspect sequence is already staggered though.

 

[HumanSteak]

About the horrors, everytimes it crashed for me was when I was fighting them. So they already did their poison clouds and were biting me as I was hitting them in close range. I suspect it has something to do with the water for some reason...

 

[Thothie]

Thinking it maybe the water myself... Sprites do fire a special event when they hit water... The horrors themselves can't reach the water (monster clip), but they may get close enough to cause that shimmering "monster in water" effect, which can spam the server quite a bit, and both their spit and spore cloud sprites touch the water.

One shortcoming of the water theory though - why wasn't this happening before? :/

 

[TheOysterHippopotami]

One shortcoming of the water theory though - why wasn't this happening before? :/

I do recall frequent crashes in the 'blue' side of lodagond-4 before the May patch. I do not recall how far back these crashes go, though.

 

[BioRage]

One shortcoming of the water theory though - why wasn't this happening before? :/

I do recall frequent crashes in the 'blue' side of lodagond-4 before the May patch. I do not recall how far back these crashes go, though.

+5.

Ran it twice, crashed, and server(s) DID NOT RESTART on LOD-4.

 

[Keldorn]

Getting a weird error from nightmare_thornlands on Blasto's server, crashes my game before I can even load my character.

 

[Keldorn]

Tried loading nightmare_thornlands on an RKS server, worked just fine. Will try loading on Blasto's again once it's empty.

 

[cartman-2000]

Steam validation error? keldorn. I've gotten that a couple times, but it only happens if I idle on the character selection screen.

From my earlier post in this thread it does look like the server might be thrashing the cpu cache a bit, loading a lot of stuff from ram which is causing the lag. So it's probably cpus that have smaller cpu caches less then 2-4 mb that get the cpu spike lag spikes.

from my amd 64 3500+ comp(2.2ghz, 512k l2 cache.)
http://imagehost.cartman-2000.net/image ... 2046E0.jpg

 

[Keldorn]

Nope, crashed the second I got to the character screen before, with some sort of write fail error.

Tried loading the map on Blasto's again, worked just fine this time.

 

[Thothie]

Sometimes test chars over 1000hp will do this when they hit the char load screen on a central server, and I guess you did recently hit that mark. Tis a rare thing though, not sure what causes it, as I don't think the HP should be loaded until you actually click on one, but it maybe some oddness with the data stream.

Hrmm... Seems I've gotten to the point where I have to put out a mini optional JUN2010 patch now, or hold off until July, as most of the changes I'm looking at now would involve client changes that would make the patch non-MAY2010a compatible.

 

[J-M v2.5.5]

I don't think the HP should be loaded until you actually click on one

I don't think so; I tested this on sfor. I joined with one Steam account, but didn't load a char, and then I joined with my main from another computer (873 HP). A certain door leading to a certain boss, requiring an average HP of 500 or more, did not open, even though only one active char (my main) was in-game. The char selection menu was still showing on my other computer.

 

[Thothie]

Things work a bit differently with FN transfers, but that trigger doesn't count up hp until you actually touch it in anycase, and only counts players that are "alive".

 

[J-M v2.5.5]

I touched it and it didn't open the door. And I was the only char alive (assuming that "Loading..." is not alive).
Or I'm completely missing what you're trying to say.

 

[Thothie]

Hrmm... (Looks at code)... The trigger demands the average hp be 500 - so if you couldn't get through with the other character stuck in load then... I guess it is loading HP before the character spawns... Very odd... But it is true, now that I look at it, I guess it isn't checking if the char is alive (guess I confused with getplayers func)... It *might* be counting it as 0hp though. Would need to test.

PS. I.... What?

 

[J-M v2.5.5]

Hrmm... (Looks at code)... The trigger demands the average hp be 500 - so if you couldn't get through with the other character stuck in load then... I guess it is loading HP before the character spawns... Very odd... But it is true, now that I look at it, I guess it isn't checking if the char is alive (guess I confused with getplayers func)... It *might* be counting it as 0hp though. Would need to test.

That's what I was trying to tell you

PS. I.... What?

 

[Keldorn]

Sometimes test chars over 1000hp will do this when they hit the char load screen on a central server, and I guess you did recently hit that mark. Tis a rare thing though, not sure what causes it, as I don't think the HP should be loaded until you actually click on one, but it maybe some oddness with the data stream.

Hrmm... Seems I've gotten to the point where I have to put out a mini optional JUN2010 patch now, or hold off until July, as most of the changes I'm looking at now would involve client changes that would make the patch non-MAY2010a compatible.

It happened to Etnies' character too, which had somewhere around 400-500 hp, so I'm not sure if it was related to my character's hit points.