Guild Wars 2

I know I will almost certainly be crucified for this, but… y’all (including the OP) are missing the point. If you want to kill a handful of players with your condi builds, go to sPvP. What differentiates WvW from sPvP is the large group (and by “large”, I mean "more than an sPvP team could hold) fights. In such fights, the conditions getting thrown around are ridiculous these days; whichever side has more condi builds will win, period. This is the issue that needs to be addressed, and if roamers get the short end of that stick, so be it.

Then again, changes to the game mechanics have always been unhealthy to WvW in that they all favor numbers over everything; I guess having condition spam added to that happy family was inevitable. As such, the chances of A’Net doing anything about anything stated in the WvW subforum are pretty much zilch…

No one competent in large scale combat uses condi builds.

That may or may not be true, but ask yourself this: How many players in this game are actually competent? All I see in group fights is everyone has 30 stacks of every condition on them at all times and if you don’t have a revenant with you to give you resistance (also at all times), you’re doomed. Maybe that means the people spamming those condis are incompetent, but it sure seems to work well for them. If you have people spewing AoE condi stuff in all directions, what’s a melee player supposed to do? We all know how this ends, namely in that “pirate ship” nonsense, and I would sure hope that nobody wants to re-open that can of worms.

In before people tell me to “L2P” or that running more condi cleanse should take care of it. Yeah, I’m sure you’d all love for us to replace all of our skills and traits with anti-condi ones, so we have even less chance of surviving the people in your condi groups that do run the occasional power build (like a hammer revenant spamming 45k Coalescence of Ruin and Phase Smash from max range on a 2.6 second cooldown, one-shotting five targets each time) and we have even less means to actually attack you back.

As I see it, the problem is this: To recover from a strike attack, you need to heal. To recover from a condition attack, you need to heal and you need to cleanse conditions. If that doesn’t say “unbalanced”, I don’t know what does. Do rebut this; I’d really like that.

The code should work every time, but sometimes the mails take quite a while to arrive or don’t arrive at all (which is a general e-mail issue, I don’t think A’Net can be faulted for that). I’d say if you get “wrong code”, is it possible you tried multiple times and the mails just arrived out of order and/or some got “swallowed”? In other words, are you sure the code you try is actually from the mail corresponding to the login attempt in question, as opposed to any other attempts you might have had?

As for the “every single time” thing: Back in the day, you could authorize individual subnets, such that you wouldn’t have to do the e-mail thing when logging in “from home”. It seems they nixed that when they changed the mechanism (from “confirmation link” to “authorization code”), so I’m not sure if that’s still possible. Either way, the e-mail verification is generally a good thing, since a lot of people publish their e-mail addresses all over the internet and, from experience, a lot of people use rather pathetic passwords. Without the e-mail verification, someone might actually guess your login credentials and take over your account. (I’m told there’s a lot of computers (many probably taken over as well), each putting out a lot of faux login attempts per minute…) Imagine logging in and finding all your gear and money gone… I’d say the protection is worth the extra effort when logging in. I mean, seriously, how many times a day are you going to log in, right? ^^

Is the 64-bit client faster than the 32-bit client?
Yes and no. While the 64-bit client can use 64-bit integers in single-step computations, the 32-bit bit client would require four-step computations to do the same thing. Being able to do in one step what the 32-bit client needs four steps for, the 64-bit client is technically faster, but the fact of the matter is that very few parts of the game actually use 64-bit integer computations. As such, the actual “bittedness” makes very little difference. However, 64-bit processors have much more internal memory (called the “register file” in technical terms) than 32-bit processors and thus a 64-bit software will run a lot faster than its 32-bit equivalent (since this “internal memory” can be accessed a lot faster than “normal” memory) for reasons that have nothing to do with actual “64-bittedness”.

What’s with the constant crashes on the 32-bit client?
The “bittedness” imposes an inherent memory limit on the software. For a 32-bit program, the memory limit is 4 GiB. For a 64-bit program, the limit is 64 EiB (in other words, “virtually unlimited”). The issue with GW2 is that is uses a lot of memory, even in the 32-bit version. Thus, the 32-bit client is liable to crash a lot because the client tends to attempt to use more than 4 GiB memory nonetheless. This issue is resolved in the 64-bit client, since this “low” memory limit doesn’t apply there.

Wait, I thought 64-bit processors were so much better than 32-bit processors…?
Not really. While being able to perform 64-bit integer computations in one step is, strictly speaking, a boon, the fact of the matter is that 95% of all software doesn’t even do this and the remaining 5% don’t do it very often. While the large register file is technically the cause for faster execution, this improvement is ultimately mitigated into nothingness by call depth: The larger register file must be saved into “normal” memory (which obviously takes longer for a larger register file) when a different function is called (which happens extremely often in high-level programming languages, such as C++, in which GW2 is written), so for any “real-world” software, the performance gain is actually negligible once the call depth limit is reached. As a matter of fact, the only reason 64-bit processors are faster than 32-bit processors (regardless(!) of whether they execute 32-bit or 64-bit software) is that they have more cache (which is memory “less internal (and thus faster)” than the register file, but “more internal (and thus faster)” than “normal” memory) than 32-bit processors.

So wait, if 128-bit processors roll around, things won’t get faster because of that?
Nope, not really.

(I don’t mean to necro threads, but since it’s sticky, I kinda figured it didn’t matter…)

What I don’t understand is why everyone keeps saying “we all knew this was coming”. Yes, we did. So? What’s your point? You really think A’Net would have been dissuaded from pushing this change though if we had complained earlier? I have news for y’all: Santa Claus isn’t real.

And no, people are usually not happy with change. But then, why do it? Ever hear the phrase “never change a running system”? I mean I understand the need for changes per se; please do fix all the bugs and do keep providing new content. But IMHO this is a change of unprecedented (and unpredictable!) gravity. Don’t pretend to know how this will impact gameplay etc., because only time will tell. What I do know is that the original sentiment of “nice way to break half our builds” is true, except it’s more like “almost all of our builds”. Whether this is for better or for worse gameplay-wise, I think people have a right to be upset (I’m positively furious right now, as evidenced by the fact that I’m even on this forum); this whole thing is tantamount to yanking the ground out from underneath us. People don’t like change? Well, guess what, we just got a huge one.

I do suspect that the impact economy-wise will be Lovecraftean, and I also suspect that this is the actually intended effect. I have somewhere around 25 characters, am I to spend thousands of gold on new builds now? And oh hey! Along comes gem to gold conversion, ka-ching!

sigh After another raid I had to abort due to ridiculous lag; time for my “from what I can see” update…

Gambit.8425:

Since that day I turned off the router mode, I have had no lag outside the “normal” WvW zerg friday reset that everyone gets. I’m no network engineer and don’t know why only GW2 was affected by this, but it would explain why VPNs can reduce the lag for those with otherwise stable connections.

It would? I don’t see how. If your router were borked, any VPN connection would be, too. And if the problem were on your end, why would the entirety of EU players have the same problem?

Zanktus.9821:

I talked to a lot people from germany and in most cases it was Telekom or another company that uses the Telekom cables/lines, for example 1&1.

That would be because, despite everyone seemingly disputing this fact, Telekom is the only real ISP in Germany. Everyone else ultimately uses their stuff. (Bla, bla, cue everyone screaming “Not so!” etc. etc.) So essentially, Germans have it bad, period.

After relentless pathpinging of the route (yes, the route — from what I can see, I (and I suspect many Germans as well) seem to get the very same route every time), I concede that my previous post was not entirely correct, either. (But we’re nearing the thuth, step by step!) The DTAG-TRANSIT15 node (80.157.129.70) does appear to fluctuate its packet loss over the day, so it seems it’s supposed to echo as well. However, my observations lead me to believe that my prior claims were not unfounded, either (Hey, I’m not just spouting nonsense, here…) — the latencies of the connections GW2 maintains seem to be fairly uncorrelated to the lag. In fact, the latency times actually dropped during lag spikes quite some times. Most of the time, however, they remained the same, lag or no.

As such, I revise my hypothesis once again: I am now convinced that there are actually two issues: A network lag problem and a server lag problem. This would explain why no proper results seem to have emerged, despite there being an absurdly huge amount of trace/ping/whatever logs available: Everyone assumes one problem and, unsurprisingly, gets contradictory results (including me, hence my twice revised theory), because there are really two.

However, I still believe (cf. lack of correlation between latency and lag, mentioned above) that server lag is the far greater of the two issues. So I have to ask: Did anyone at ArenaNet/NCSoft actually conduct any tests to rule out my bold (though well-founded) claim that the servers are running out of resources? Like I suggested before, bypassing the internet and connecting to the servers “directly” would quickly show you one way or another.

Y’know, because pursuing the network issue might be a waste of resources if the server issue really is the far greater one… And, as I mentioned before, do consider the fact that there was no lag (in comparison) not too long ago — the assumption that the internet got borked a few months back seems far more outrageous than the assumption that the player base increased and someone neglected to buy more servers. (Occam’s razor, anyone?)

And yet another WvW raid foiled by terrible lag and constant disconnects. This is getting ridiculous. 2015-03-02 around 22:00 CEST. You really need to buy more servers. Network congestion does not account for multi-minute freezes, especially when the people have vastly different routes.

I know that license agreement states that you can do whatever you want and we are not allowed to complain, but you’re not helping your image here.

OK, with tonight’s findings, and observing the flow with all this in mind, I’m more and more getting the feeling we might have gone about this all wrong.

- The VPN tunnel to the US didn’t resolve the issue, despite altering the route to a different T1 as intended.
- There were times when I could play fairly decently with the tunnel, despite having 70% (!) packet loss the entire time.
- If somebody could name a server in Texas to which I could maintain a TCP connection for an extended period of time, I would do so and check whether I get disconnected from it when I get disconnected from GW2. I do get the feeling I would not.
- The lag issues ramp up in the evenings, when most people play GW2. Someone mentioned that WildStar (also by NCSoft, no?) apparently has the same issue — probably also played in the evening, when people come home from their jobs…
- It used to work flawlessly, but has been slowly getting worse and worse. Now consider that different ISPs, including different T1s, are in use and are all experiencing this issue. I refuse to believe that, starting 4 months ago, the backbone networks have started messing up the (entire!) internet and keep making it worse.

As such, I am now convinced that the issue is, in fact, server lag rather than network lag; this would, IMHO, explain all of the above. ArenaNet/NCSoft have been selling more and more copies of the game and the lag has been getting worse and worse. Just recently, the whole “75% off GW2” thing happened and, sure enough (for me, anyway), the lag had a quantum leap. Did you, by any chance, forget to buy new servers to accomodate the ever-increasing player base? Did you perhaps increase the population limit of the existing servers to offset it? Some of our guild literally “passed on” their lag to other players tonight… I smell insufficient resources on the servers.

I assume you ArenaNet people have tried to reproduce the problem? If it was me, I would now take my laptop right into the datacenter and plug it into the local network there, then “play” on an EU server in the evenings every day for one week. That should quickly determine whether the internet is the issue. If the lag persists despite the “direct connection” (which I’ve come to suspect), you’ll know for sure (and the vast majority of posts in this thread (including my own) as well as the “e-mail the logs” campaign might be rendered moot).

Edit: I realize that my revised theory effectively contradicts my previous one. However, when I said “If the problem were with NCSoft, everyone would have this problem, but only EU seems to have it”, I assumed equal server load between NA and EU. If you drop this overzealous assumption, the NCSoft servers being the problem becomes quite conceivable indeed.

Horus Nightlight.1276:

UncleSniper… If you care to look at the the screenshot I took of the pingplotter which I have attached in the post above yours, it is proof that the problem isn’t with the local servers. And while it is true, there is lag in the backbone nodes, it is the only at the NCSoft IPs where there are packet loss of up to 28% (10% first time round and then 20% next time). While I am no IT expert, I believe that packet loss only occurs in 2 settings -

1. Corrupt data
2. Congested network (more common).

Looking at the picture and performing whois on the nodes there, I get:

192.168.0.1 = IANA private, obviously = “me” in my above scenario
115.132.109.* = TELEKOM MALAYSIA BERHAD [MY] = “my ISP”
10.55.32.148 = IANA private (WTF is that doing in there, anyway?)
all the .telia.net nodes = TeliaSonera AB [EU] (duh)
all the .ncsoft.com nodes = NC INTERACTIVE, INC. [US -> TX] = “NCSoft”

Since there are no nodes between the .telia.net and .ncsoft.com blocks, I must assume that NCSoft is directly on Telia, so those nodes equate both the “backbones” and the “NCSoft’s ISP” parts in my scenario. Now looking at where the latency spikes are introduced on your graph, that’s pretty much exactly in the Telia nodes — therefore, my theory of “NCSoft’s ISP is borked” is actually supported by your log.

As for the packet loss, the ping packets are very likely routed exactly as the trace indicates. That means, in theory, any node “further up” could be dropping them, such that they don’t arrive at (or return from) the node in question. In other words, it’s quite concievable that Telia is the one losing the packets. If it were NCSoft’s servers, that would mean the EU servers are dropping lots of packets, but the NA server are not — and yet, both seem to be in the same datacenter, making this interpretation extremely unlikely.

Edit: To be fair, the point where the latency spikes is essentially the point where the transatlantic lines sit, so delay in principle is almost inevitable at that point. However, since no bigger spikes are further down the pipeline, we must assume that the spike that is there is actually the issue. As such, my theory still holds; consequently, NCSoft needs to step into Telia’s toes.

Edit 2: On second thought, the change in continent would be further into the block, whereas the increase in latency is not. Perhaps disregard edit 1. :P Trying tunneling via USA for tonights raid — we’ll see whether this solves things. Either way, I got the pathping thingy all primed and ready in a batch script, so if massive lag strikes again, I’ll at least have my own trace to fall back on.

This lag issue is getting worse and worse every day. Last night, I literally had like a three minute freeze of the game. After that, the events flushed though and I got to watch myself down, rally a bunch of times, eventually get defeated and revived, all within a few seconds. About half a minute after, it froze again. I think it should be obvious that this qualifies as “unplayable”.

Now, there have been lots of people throwing around their opinions as to who is at fault here, but (merely skimming this thead due to the huge number of posts) I get the impression that logic is rarely involved. So let’s look at this scientifically: Correct me if I’m wrong, but the connection goes somewhat like this:

me >> my ISP >> backbone types nodes >> NCSoft’s ISP >> NCSoft

If the issue were with…
- …me, this thread would not exist, as I would be (almost) the only one having this problem.
- …my ISP, at the very most, only Germans would have this problem, but it seems to concern all of Europe.
- …the internet backbones, we could tell by the all around pillaging and plundering, since the internet breaking down is tantamount to the apocalypse in this day and age.
- …NCSoft, everyone would have this problem, but only EU seems to have it with NA working fine.

This leaves only one: The networks on NCSoft’s “end” seem to have issues. Now, many of the traces have ostensibly confirmed that this is the case. With the number of pathping logs posted here and Haruhi knows how many emails with such logs sent to the support, I would think there should be enough data to root out the network(s) causing problems. That said — how to put this delicately? Get a proper ISP, NCSoft! And, with that down, drop the one/s that is/are garbage. Clearly, some of the networks (through which EU traffic seems to be routed, but not NA traffic) are about as good at providing connectivity as a blind carrier pidgeon.

Soooo, you’re only stating what happens to the “big” events (world bosses, temples and whatnot), namely a fixed schedule. But what happens to “normal” events? On a bunch of maps, there is the practice of running down champions and it is nice to see which champs are currently up, e.g.: http://api.unclesniper.org/gw2/events/status/ros/queensdale-train.xml
Are you really going to drop that entirely? That would be really sad, especially since, as has been mentioned before, it would (in theory) still be possible to expose all event status — only by instance ID rather than map ID. Any chance we will get this functionality back at some point?

As for the message pasting, who knows how much of what you do the client communicates back to the server? A’Net (who can obviously program at a rather high level, too) could easily have automated detection of such things.

As for the “unauthorized distribution” of tools, them catching you is just a Google search away.

I’d much rather have them tell me it’s OK than getting my account terminated (or getting sued, respectively) over nothing.

Hello, everyone; I have two questions regarding the intention of the Rules of Conduct with respect to third-party tools. I have no idea whether this is the correct sub-forum for either of them, but after checking the list, this one seems like the closest match to me.

First question: There routinely are exchanges like so in the game:
Alice: anything up?
Bob: champ something-or-other is up [Whatever Waypoint]
What if Bob, after checking some list of active events (based on the A’Net server API), copy-and-pasted a pre-made message (i.e. the “champ something-or-other is up [&weird-base64]” part) into the chat? Rule 20 states that “You may not use any third-party program […] in order to automate gameplay functions, including […] chatting”, but manually alt+tabbing over to the list, ctrl+c’ing the message, alt-tabbing back to the game and enter-ctrl+v-enter’ing the message is hardly full automation, is it?

• I know they say “one button, one action”, but what counts as an action here? Is inserting a single character into the text field really an action in itsself? The very fact that ctrl+v is even supported in the chat seems to indicate otherwise…
• You’re not supposed to gain any advantage; in particular, nothing allowing you to do things “faster or better” than anyone not using the program in question. I suppose one could argue that pasting the message is faster than typing it, but the “program in question”, here really is Windows itsself (allowing the copying of text) and everyone is using that, right?
• I suspect the “no automating chatting” rule is intended to prevent spam/scam bots from littering the chat with unsolicited/malicious messages, but something like the above is neither unsolicited nor malicious, is it? I realize making exceptions to the rules cannot possibly be expected, but due investigation is required before dispensing punishment anyway (right?), so it’s really merely a matter of not banning people when it is obvious that no harm or unfairness was intended…

In summary: Is someone copy-and-pasting such a message liable to get banned or not? IMHO, all reason would suggest that they are not.

Second question: Rule 19 states that “you will not post or distribute any utilities, […] or other software tools related to Guild Wars 2 without the express written permission of ArenaNet or NCsoft.” Looking at the “List of apps/websites using the API” thread in the “API development” subforum reveals a huge list of people that do distribute their own tools, which clearly “relate to GW2”. Am I to understand that all of these authors are in possession of a physical piece of paper proving that A’Net allows them to do so? Somehow, I seriously doubt that. As such, am I correct to assume that rule 19 is intended as a “no endorsement” kind of thing to prevent people from trying to make money off of A’Net/NCsoft trademarks? That is, if I were to write a utility for use with GW2, would I really need to apply for such a slip of paper just to make it available for public download? That would seem extremely counter-productive, if you don’t mind me saying.

Kind regards,
UncleSniper

Using the “target nearest enemy” button is something I for one do all the time. Now, if there is a “supply stack” (passive/yellow) in the vincinity and a “Champion Corrupted Rapaging Mauling XXX Chieftain, Level 999” (aggressive/red) charging at me, which do you think I want to attack right now? The game seems to think the former. The controls are hard enough to work as it is (circle strafing and targeting skills at the same time?), but that’s just plain annoying, having people repeatedly click on monsters to target them.

Kind Regards,
UncleSniper