changes on test that will effect live servers soon

[kunark]

"Changes have been made to the way the client is notified about other entities (players, NPCs, etc.) in the zone. Watch for any problems related to zoning, teleporting, and players or NPCs coming in and out of your actor clip range."



Is that anti-MQ code?
It's more a secondary aspect of the changes that prompted the feedback on performance. It's not specfically anti-hacking, though there are some aspects that will limit the usefulness of some 3rd party apps.
To explain a bit more, the client doesn't really need to know about the whole zone at any particular time. It needs to know about what is around your PC and areas your PC is likely to move to next. Everything else is basically wasted information so this change is to not send the information the client doesn't need to know about. By itself, this change won't do too much, but it ties into other changes we're making which could have a significant impact on performance.
Rashere



these will go to live when they get bugs worked out.
so how bad do you think this will effect us, or will it kill the program?

[Azarkon]

Yes, they will.

All entities outside of a certain radius are no longer being sent to the client. That means you can't detect them, period. Trackers might be given a larger radius than non-trackers, but overall it's the end of ShowEQ as you know it.

[BlueAdept]

I hope not, but it does sound like it is directed at SEQ and not MQ. I bet that names will still be sent, just no positions.

Oh well. It was a good 9 years. If it is true, it will probably be the end for me. I was planning on leaving anyway. I had quit once already and I really do not have time to play anymore. My renewal is coming up again and wasnt planning on renewing but this I think will definately be the last nail.

[Seaxouri]

Actually they have already been doing that to a point. There is a radius around the player that gets updates. When you first zone into the zone, you get a list of all spawns in the zone, but you may not get position updates for mobs across the zone. What I had noticed was that if you have a someone in your group or a spawn targeted, you will get updates for that target for a longer range. However at some point the spawn appears to freeze. They really are not freezing, but rather going out of range.

I have also seen this when spawns appear to be stuck far away from you. Imagine if you are at zonein along with a few other players and you start heading straight south in a large zone. Now imagine the spawns at the zonein were spinning in a circle. You would see them spinning when you were close to them. As you move away from them, at some point they will stop spinning and freeze in place. Here the server has stopped sending you information on those spawns, but there is still a placeholder in memory for them. Now imagine those spawns get COHd to a spot even further south than you are. You will still 'see' them North of you at zonein. Not until you keep moving south and get close to them again will the spawn update and the dot appear to 'warp/jump' to the new spot in the south part of the zone. This behaviour has been active for 2 years now at least.

What they could do to kill SEQ would be to:
1) Remove the placeholder from memory if you get too far. This would mean you only see spawns in a radius that you can see in game anyhow.
2) Decrease the radius of the spawns they update. It would work like it does today, but spawns would 'freeze' in a much tighter circle.
3) Only send spawn info for mobs within visual range. This would act more like #1 above at zonein and then as you move around the zone, fill in more spawns, although ones that go out of range might freeze.

The methods they are using, or could use, are more like what WoW uses. As you move around, mobs appear and disappear. One thing that would be affected would be tracking. They would need to maintain a separate list of tracking info for each spawn, but that could be a small memory footprint and not contain enough info for MySEQ to track all trackable spawns. Also, they could shut this off for non-track classes making MySEQ only somewhat usable for a few classes.

It certainly would make the game run a lot smoother in large zones. For bards running fast, though, it certainly might cause a lot more server lag, since the spawn engine would be allocating and freeing a lot more frequently.

Whatever happens, it is possible that the end of MySEQ is coming. Personally I think the end of EQ is coming.. at least to the point where Sony will sell it off to some company for a chunk of change and get out of it altogether, and focus on the next generation MMOGs. EQ subscriptions have seen their peak and been dropping for the last four years.. most sharply in the last year. EQ is far behind many other games you probably have never even heard of. Subscription levels are back down to where they were in the Summer of 1999.. just 3 months after opening their doors. In January 2008, there were less than 150K subscribers.. and that number has been dropping at the rate of 100k per year for 3 years.

I would not doubt that we will see an annoucement for them to raise prices very soon to compensate for the loss in customers. This will be the final nail in the coffin because customers will not renew at the new higher prices.

[rasthan]

The thread:

http://forums.station.sony.com/eq/posts/list.m?topic_id=130109

And a summary of what I read on that thread:

All data will be sent to the client on a need-to-know basis.

The need-to-know basis will consist of your client's visual clipping setting. When you zone, only the data for the npcs/players/objects in your visual clipping range will be sent. The visual clipping setting will be hard capped on the server, for data-sending purposes, in case someone is running a plugin to extend their client's clip beyond the norm.

Players in your group and whatever you have targeted will be sent to your client, no matter the location.

Tracking classes will have a list sent to their client by the server, which will update every six seconds. The list will not auto-update if you don't want it to. Location information is not sent with this list until you pick something on the list.

The dev claims this change is not aimed at squashing third party programs. SOE has long had a space problem with "global models". The present client pre-loads the models for every npc/player/object inside the zone. The second phase of this change involves dynamically loading textures the way modern MMOs do. This way, the client won't need to load textures for everything in the zone. Only for when you get near something.

It will pave the way for having new looking pets, mounts, etc.

Thats it for now.

[Seaxouri]

Could pave the way for seamless zoning too.

The model data makes a lot of sense to me. That is pretty dumb to load all the textures for all mobs in the zone, considering you may not see 80% of the mobs as you pass thru the zone.

In any case, the amount of memory used to hold 'basic' spawn information is very, very small. It is still possible they will maintain a complete list of spawn information, but only send position/speed/direction updates for those spawns in range and load texture information for spawns in visible range.

Part of the problem is that if they wait to send *everything* across the net, thats a lot more data that needs to come across the wire as someone runs thru the zone. Imagine you running in a large circle, as you run, spawns will pop in and out of your radius. If you freed everything that left the radius and alloced everything that entered it, you would be freeing and allocing a LOT and each spawn that popped in and out would have to be sent across the wire.

My point is, that to make it work somewhat smoothly, there needs to be *some* data cached or kept in a much larger radius than you can see to prevent this thrashing. In addition, there needs to be some sort of hystersis in the algorithm so that there is no magic spot where you step left or right one step and cause a ton of mobs to pop in/out of your circle. Much like the way your thermostat works in your house. If you set it to heat/70, it wont kick on until the temp drops below 68 and off until it reaches 72. So it has a hysteresis range of 68-72. If it were hard set at 70, then a slight cool or warm breeze would trigger it on/off, and that would thrash the system.

[Carpathian]

Looks like its, for the most part, game over guys. Here is a picture of how MySEQ looks running on the test server. It was indeed a good run though!

http://i282.photobucket.com/albums/kk260/carpathiantwilight/eq_newchange.gif

What they're doing is only sending mob updates within a bubble radius. If a mob falls out of that bubble, the spawn is removed from memory completely. If anyone is interested in the offsets/code for the test server, I'll post it.

[Carpathian]

I tried it with a tracking class just now. Very interesting. It appears they're keeping the spawns and the tracking data separate from each other. I can see enemies on track which aren't on the map.

The next logical step would be to find the tracking information and see how much information it contains, however, I doubt its much more than the spawn name and (perhaps) a heading from your characters location.

[Seaxouri]

It would be interesting to see if they are keeping two lists. One that gets updates and the other that does not.

If someone on test can scan memory for other PC's in zone, maybe we can find a second linked list as well.

Just a thought.

[Carpathian]

I'm doing a little bit of digging myself at the moment, and will let you know when I have a better idea of whats going on for sure. I will say that there is indeed a secondary tracking list. I'm trying to find where the pointer for it is stored at the moment. Another interesting fact is that it doesn't not appear to have next or previous pointers, so this leads me to believe its not a linked list. Once I find some more information I'll let you know.

EDIT: Posted my code to make MySEQ work with the test server in the development forum for anyone that is interested. So far from what I can tell the only values each item contains is: Name, SpawnID and Level.

[Fatal]

Wouldn't tracking also have to contain the xyz of those mobs in order to correctly track them in the track window?

Hard to believe this could be the end of ShowEQ as we know it. What a long strange trip its been.

[Hidron]

Possibly for a distance based track. I don't know how often people use it though. Only use it myself for tracking things I'm farming for tradeskills.

[Carpathian]

Wouldn't tracking also have to contain the xyz of those mobs in order to correctly track them in the track window?

I thought about that question as well. However, since you can only track the location of one entity at a time, there would be no purpose for it to get the X,Y,Z until you ask for it.

[rogues]

Well, as per the patch message for 4/17 (today), these changes are going through. Really, it's such a huge change to the game back-end and client that I would have expected it to take more time for them to iron out the bugs (I should reword that "it will take a lot of time for them to iron out the bugs now that they've moved it to Live").

It's been a great trip, and I've really appreicated everybody's efforts. As somebody who spends a lot of his time pretty lost in this game prior without this, it's going to be very disappointing to see it gone.

I do think that even what Carp had is still useful to some extent, at least for myself.

[Wyntyr]

Any chance you'll continue posting the offsets? I;m a relatively new user but have grown to really enjoy some of the features, pretty sure I can live with the limitations posted above.

[Hidron]

The spawn list populating/depopulating as I run around is already getting to be really annoying.

[Hidron]

Just noticed that ground spawns still seem to show for the entire zone.

[o___o]

Is this 'range' the same as the trackers have? Even SEQ in the crippled state this change for EQ engine would be valuable to non-tracking classes, if we can get it to function to the same extent that track works for tracking classes. Beyond that, using it for a GPS would be useful as well.

If there won't be any future offsets published for SEQ users, how difficult is it to find the offsets ourselves?

[Hidron]

The patch changes seemed to break tracking too. There is a 2-3 second delay after pressing the track hotkey now before the window opens, and even with players set to off, they are periodically displaying on track, sometimes more than once in the list.

Mob range in MySEQ seems fairly small everywhere I have been so far aside from the castle area in Dreadspire Keep. Seemed to be showing everything for me in the castle as soon as I clicked over friom the zone in side. Maybe this had something to do with the map somehow though as I have to move the map (usually just move the scrollbar on mouse) to get the map to display where I now was.

[Carpathian]

Another interesting piece of information to note... when you die and use the in-zone resurrection window, all of the mobs in the zone will show up on the map. Not quite sure why thats going on.

[Oggre]

when you die and use the in-zone resurrection window, all of the mobs in the zone will show up on the map. Not quite sure why thats going on.


Interesting..

Is this a temporary or permanent (until zoning) issue? meaning, do they show up momentarily and then disappear, or are they displayed on the map and list until you leave zone? Maybe this could be exploited to find a fix.

O

[kunark]

Interesting..

Is this a temporary or permanent (until zoning) issue? meaning, do they show up momentarily and then disappear, or are they displayed on the map and list until you leave zone? Maybe this could be exploited to find a fix.

O

can confirm, died got inzone res ( never returned to bind) repoped and all mobs up, stayed up whole night till logged, also noted hidecorpses allbutgroup did not work after res.

would hate to have to die and rez in a hunt zone to work lol, hope this helps


later died and rezed worked agian, died soon after from bad pull got res ( ld'ed on clicky repop ) and when back in game map was back to normal new small circle crap. ( note: same server from first post to this one was still running, so no new memory reads as in ending and restarting it etc )

[kunark]

a note to the above post, inzone rez and repop doesnt spawn entire map each time, id say out of 5 deaths 3 popped whole zone.

another note, while whole zone was popped, i could click some grp members and it would show them on map, others would not show nor would there name show in target box.

on further testing, died agian draged self to zone line, got rezzed, all mobs popped, i zoned and map was back to small circle of mobs.

hope this helps, not dieing anymore ( by choice lol )

[Carpathian]

Seems Sony is going to be setting the radius to infinity with the next patch (for those who may not read other parts of the forums here or the station boards). Good news, even if it is only temporary... perhaps they'll just give up ;-).

[BlueAdept]

They have given up many times in the past, but I personally think this repreave will be only temporary. They want to eliminate GKing and this is the simplest solution...until they find a different way to GK.

[kunark]

Seems Sony is going to be setting the radius to infinity with the next patch (for those who may not read other parts of the forums here or the station boards). Good news, even if it is only temporary... perhaps they'll just give up ;-).

You know speaking on this line, i have noticed that while in game turning the actor clip plane up or down, also reduces or enlarges the area in map and number of mobs shown.

so my question is, ( if anyone knows ) is the info or cap for actor plane stored in an *.ini file, and if so can we increase the number or cap?

it seems like most mobs show with my clip maxed at about 600 or so feet, so if cap is say 100 in file what happens if we take it to 1k or higher ( assuming its even in a file, but i would guess it would have to be, so the client knows how far to view mobs).

anyway this is just a thought i had last night while i was posting the above post, but havent found it yet to play with.

[purple]

I'm pretty sure min/max clip plane settings come down from the sever in zone info. I'd imagine that the client is sending clip plane changes to the server now, but if the server doesn't apply the min/max for the zone on the server side as well, Sony is very stupid. You might be able to change the max in memory, but the server really shouldn't listen to you if you do that.

[Carpathian]

You used to be able to change the min/max clip plane by editing the value in memory. It's stored in the ZONEINFO struct in memory. We already use this information for the zone short/long name. I do recall them saying something on the SOE boards about limiting the max in case people did try to change the clip-plane distance. I can give someone the offset if they want to mess with it, but I'm going to leave it alone :cool:.

[Hidron]

One thing I just noticed tonight is that when I have /hidecorpse looted set, when the corpse is hidden it no longer shows on the MYSeq map. This was not the case before todays patch, or could have missed seeing it after last weeks patch, but it seems to be a new occurrance.

[Seaxouri]

They have given up many times in the past, but I personally think this repreave will be only temporary. They want to eliminate GKing and this is the simplest solution...until they find a different way to GK.

They already have a method for determining who is GKing. I know of several folks (about 8) that used MQ2 with the GK plugin and they are now banned for life. No second chances. This was done well before the clipping change, so I don't really see the connection between the two.

[BlueAdept]

A senior guide told me. He said he was told by one of the Devs.

From what I was told, yes they have had ways to tell of GKing, but it didn't prevent it. It also had to be researched by someone, I assume, since you dont get autobanned the moment you start doing it. The reduced radus was the "easy" way to prevent gking from happening now and in the future.