BUFULO's dashboard

I think my tapping position accuracy is pretty darn good. I have trouble with tapping speed on the really fast songs or if there are too many fast taps in a row. Guess that's what makes those songs Extreme though! During normal play, I don't think I would run into the overlap issue even if it did exist on my device (which I still can't demonstrate). – bufulo, on August 20, 2009 19:19

All on a PC running Windows XP. I've tried Firefox in the past & dumped it because there was just too much web content that didn't work with it, especially secure sites for banking & stuff like that. I liked it otherwise, but dang MS still has a hold on web stuff & I doubt that will change anytime soon as long as they keep bundling IE with the OS.

Since XP & IE 6 works for Pyro637, it might be a security setting issue or maybe an Add-on issue. I think I'm running all the latest for Flash & Java but I think GetSatisfaction only uses Javascript. Maybe I'll try deleting some cookies (mmm... cookies...) & reinstalling Java too. – bufulo, on August 20, 2009 19:09

Oh, here it is! Found the screenshot of the incident that prompted this post in the first place:

Here's one where the winner was definitely a little under-rated!

Finally got a way to post images (thanks Daniel Thomson for the tinypic.com tip)!

I can't find the image I mentioned in my original post, but here's a shot of dualis, MajorasMask, & Leftfield seriously getting their asses kicked: :)

Thanks, but I don't use Twinkle. So you can "Add an image" on this website & it works? What OS & browser are you using? – bufulo, on August 20, 2009 18:06

I get "TypeError: 'responseText' is null or not an object" every time. The files are tiny & PNG format. I'm pretty sure I tried JPG in the past & got the same message. Must be a PC vs. Mac thing.

OK, the "Direct Link for Layouts" on tinypic.com works!

Thanks for the help with the "Add an image" mess. I'll try a few things this morning to get them posted.

I'm curious what browser (& version) & computer OS (& version) you guys are running if "Add an image" works for you. I'm on Windows XP SP3 & have tried IE7, IE8, & Safari. Is Get Satisfaction Mac-centric or something?

I've been able to download songs via 3G & wifi, but have had trouble at times with either. I've never had songs "go missing" as far as I know though. There are so many that I'm not sure I'd notice if a few disappeared. I don't doubt this sort of thing is possible though & it would tie in nicely with some of my memory management, fragmentation, & performance degrading with time theories. – bufulo, on August 20, 2009 07:57

I can relate. Same thing has happened to me on lots of songs, especially online. I can imagine a few ways this could happen but can't prove any of them. You could try deleting & re-downloading, but this didn't help at all for me.

My guess is you're just starting to experience what most of us 3GS (or OS 3.0) users are seeing. It will get worse with time. Hopefully OS 3.1 solves these issues.

Yeah, laugh it up fuzzball. :) I can't wait for the new stuff, but I suppose it won't get out to us plebes until after Labor Day.

It wouldn't surprise me at all if the graphics updates & fixes solved some of the issues I've been discussing - latency & memory. I'm sure the graphics shares system memory & has plenty of critical (non-interruptible) sections of code. That alone doesn't explain the degrading performance with time though, so I suspect there is more to the story.

If I were you, I'd make a Backup of your current config now & hold onto it for safe-keeping! :) – bufulo, on August 20, 2009 07:47

I'm glad you guys agree, but I'm pretty sure there are issues in TTR as well. I've downgraded my Touch 2G to pre-3.0 versions of the OS & taps still dropped. My understanding is that there is some chipset firmware than cannot be downgraded though, so technically this test isn't totally valid. No argument that the arrows issue is very strongly related to OS 3.0.

This is why I think there are both OS latency issues AND memory management/corruption issues (or maybe something else that I just haven't thought of yet that can degrade performance with time). The latter could be either TTR or Apple or both - there is no way for us to know. So I suspect OS 3.1 could very well solve the problems, but only time will tell. – bufulo, on August 20, 2009 07:31

So can you explain why the targets overlap & also why I don't see any evidence of overlap?

The game is only as "smart" as the programmers writing the code! :) I'm sure they're fine programmers, but don't underestimate how difficult it can be to get all the information necessary to write a real-time application such as this! I can pretty much guarantee Apple doesn't publish interrupt & task latency figures! Without those, it's just a guessing game, but probably one that can be made to usually work until the timing is just wrong!

I'd be happy to post some videos of plenty of dropped taps if I could come up with a way to easily do it without taking too much time. I tried having my wife use her phone to take video while I played the other day, but it didn't go very well. Maybe with a little more practice I can get a few posted on YouTube or something. Trust me, it drops a lot, especially with online play! – bufulo, on August 20, 2009 06:49

The "background" tasks like wifi & 3G are probably given higher priority in the Apple OS, so they essentially interrupt & block TTR & tasks related to touch events. If written optimally, the amount of time they block is kept to a bare minimum & really shouldn't cause any problems.

My guess is that this is where one of the major problems rests & Tapulous can't do a darn thing about it. Hopefully OS 3.1 will be better optimized. Ideally touch related tasks could be temporarily bumped up to be higher priority than network tasks, but I seriously doubt that will ever happen.

Playing with those tasks disabled (at least to the extent we can) definitely helps A LOT! I agree though, it totally defeats the purpose if you can't Submit your score or play with others. What fun is that hey?

The "getting worse over time" issue is really bizarre. When I first got my 3GS, I didn't even bother using Airplane Mode, turning off Push, or Auto Brightness because it really didn't seem to make a difference. It sure does now!

The only explanation I can come up with is memory management issues - either fragmentation or corruption. This is either causing some background memory management task (probably OS) to take more time in a critical section of code (blocking as discussed above) or is actually outright memory corruption or leakage (memory allocated but never returned to the system) by either Apple or TTR or both. I've seen behavior that could be evidence of any of these things, but there is no way for us to debug such a thing. This is something Apple and/or Tapulous have to figure out.

PS. One thing I've often wondered is whether or not the hold scoring is timer driven or just in a loop in the code (TTR). If it's not driven by a timer (interrupt), your score may increase more slowly if background tasks interrupt TTR! In other words, if Tapulous didn't think of this, it may very well be possible to get your best scores with as many background tasks as possible disabled! :O – bufulo, on August 20, 2009 06:40

Yes, I do tap only with my thumbs & they are pretty big, but I actually turn them down to use the tips. I've found that if I turn them down too much to use the very very tip, besides cramping easily after playing alot, the taps don't register quite as well. My guess is they are slightly less capacitive in that area due to reduced blood flow (salt, conduction, whatever). Minor digression. FWIW, I WOULD prefer higher tap targets! My thumbs start to cramp after playing an hour or so the way it is now.

Anyway, I messed with Night Falls as you describe & I can only get it to register one tap at a time when I tap in-between. Maybe I'm still missing something, but you are making contact with both targets at the same time, right? It's just not possible to tap only in-between unless your fingers are as narrow as toothpicks! :)

I didn't try A Fact of Life, but your description sound like exactly what should happen. If you tap anywhere in the left target it should register in that lane only & drop the middle as you see happen.

So the tap target area doesn't overlap at all for me & that makes perfect sense. In fact, there should be a few pixel wide strip between each lane specifically to ensure you can't do exactly what you describe (tapping in-between & having it register in both lanes)! This just seems like common sense. I can't understand why on Earth Tapulous would have the tap area overlap. They should just compare the center of the touch to the tap target bounds (the colored circles). If the center of the touch is on or within those circles, it counts for that lane. Piece of cake! Or is that chicken soup? – bufulo, on August 20, 2009 06:22

First time I've seen it. I have seen the white blocks a few times.

I should probably quite posting by dissatisfaction lately. I broke 4 ribs in May & they're not healing so I've been pretty darn crabby lately. – bufulo, on August 20, 2009 01:28

Ha, this is fun. I'll "TRY AND BOTHER" krohn7master! :)

I've seen lots of obviously dropped taps playing easy songs (like Hey You or Contagious) on my 3GS with PLENTY of free RAM (over 150 MB). Of course the only way I can verify free memory is with an app outside of TTR, so if the OS or TTR is blowing all the memory while TTR is running, there would be no way for me to know. Regardless, that's WAY more memory than a 3G or earlier device would have free. Upon exiting TTR, I've noticed no memory leaks either. I've also seen no correlation whatsover with lag/stutter & dropped taps.

Are you POSITIVE there is overlap in the valid tap area? That does not match the behavior I see on my 3GS at all! Some part of my finger MUST hit the tap target graphic in the proper lane for it to register. Even if this is truly the design for some "clever" reason, why not get rid of the overlap & solve the problem?

Also, your description of the OS touch recognition is sort of incomplete - simplified to make your point seem legit. Each tap is forwarded to the app with timestamp & area touched. Any subsequent movement is also forwarded, at the application-specified rate, but those events SHOULD just be ignored by TTR. There could be some complications I'm misunderstanding, but I don't think so or at least nobody has offered anything convincing yet.

I don't disagree that most of these troubles probably originate in Apple OS, but definitely not all of them. – bufulo, on August 20, 2009 00:37

So what song(s) did you use to come up with this "overlap" theory? I just messed around a little & don't see that behavior at all.

The tap targets don't seem to overlap the slightest on mine. I tested this with 10,000 Miles or whatever on Easy. Some portion of my finger MUST contact the appropriate tap target for it to be recognized.

TTR gets notified when a touch event occurs. The notification includes a data structure containing a timestamp & touch bounds info. TTR then probably compares the timestamp & touch bounds to the potential target bounds & current tapper time. Seems pretty simple to me & seems to be exactly how it works for me (iPhone 3GS).

I still think the trouble is with other OS tasks adding latency & probably even some task or interrupt priority issues. Touch events need to be pretty high priority for this game to work well but I'm not sure Tapulous has any control over that!

I don't understand why the valid tap areas would overlap at all. That sounds like some software engineer trying to be clever but actually opening up a big can of whoop-ass... I mean worms.