3.0 breaks SoundEngine.cpp

[Bertrand21]

/Users/rbertrand/Desktop/Iphone Projects/fakesms/Classes/SoundEngine.cpp:706: error: jump to label 'end'
/Users/rbertrand/Desktop/Iphone Projects/fakesms/Classes/SoundEngine.cpp:682: error: from here
/Users/rbertrand/Desktop/Iphone Projects/fakesms/Classes/SoundEngine.cpp:686: error: crosses initialization of 'OSStatus err'
/Users/rbertrand/Desktop/Iphone Projects/fakesms/Classes/SoundEngine.cpp:706: error: jump to label 'end'
/Users/rbertrand/Desktop/Iphone Projects/fakesms/Classes/SoundEngine.cpp:679: error: from here
/Users/rbertrand/Desktop/Iphone Projects/fakesms/Classes/SoundEngine.cpp:686: error: crosses initialization of 'OSStatus err'
/Users/rbertrand/Desktop/Iphone Projects/fakesms/Classes/SoundEngine.cpp:706: error: jump to label 'end'
/Users/rbertrand/Desktop/Iphone Projects/fakesms/Classes/SoundEngine.cpp:670: error: from here
/Users/rbertrand/Desktop/Iphone Projects/fakesms/Classes/SoundEngine.cpp:686: error: crosses initialization of 'OSStatus err'
/Users/rbertrand/Desktop/Iphone Projects/fakesms/Classes/SoundEngine.cpp:761: error: jump to label 'end'
/Users/rbertrand/Desktop/Iphone Projects/fakesms/Classes/SoundEngine.cpp:721: error: from here
/Users/rbertrand/Desktop/Iphone Projects/fakesms/Classes/SoundEngine.cpp:722: error: crosses initialization of 'bool isFormatVBR'

[scotopia]

Dang; my apps use this engine... If they are still built for 2.0 or whatnot, will they crash on an iPhone running 3.0; or is it only that you cannot build it for 3.0?

[Bertrand21]

No it works if you build with 2.x.x but building with 3.0 kills it

[tawpie]

Originally Posted by Bertrand21

No it works if you build with 2.x.x but building with 3.0 kills it

get we had better get to the redesign... splains why crash landing was pulled.

[wuf810]

Originally Posted by tawpie

get we had better get to the redesign... splains why crash landing was pulled.

and why they implemented AVAudioplayer etc. and also why I never used SoundEngine...been bitten too many times by code that suddenly is nolonger supported.

[scotopia]

Originally Posted by wuf810

and why they implemented AVAudioplayer etc. and also why I never used SoundEngine...been bitten too many times by code that suddenly is nolonger supported.

Yeah I had a sneaking suspicion this was why the soundEngine.cpp was yanked... oh well. But if it still works on 3.0 iphones when the app is built for 2.0 (or whatever) it's not terrible, as it won't "break" anything already made correct? We'd just have to change it if we want to build for 3.0; is my thinking on target?

[Delirium]

Guys, I have SoundEngine working in the 3.0 simulator. I do not have the OS on a phone, so I can't verify that. Both of my games use SoundEngine, and like others in this thread, I got the same errors when compiling for 3.0. However, all you need to do is replace all the AssertNoError calls with:

if ( result != noErr )
return result;


and then remove the end: line.

Once you do this, it will compile, and run just fine in the simulator. I'm getting full audio, with no problems. Would anyone who is running 3.0 on a device be willing to replace these lines and try it out for themselves?

[tawpie]

Originally Posted by Delirium

Once you do this, it will compile, and run just fine in the simulator. I'm getting full audio, with no problems. Would anyone who is running 3.0 on a device be willing to replace these lines and try it out for themselves?

Delirium...

I finally got around to incorporating your fix and for a device build, the compiler complains about returning a value (result) in a void function which is easy enough to handle.

Before I spend too much time experimenting, it would appear that the AssertNoError function is there to print an error message, and then skip merrily along to the "end:" label/line so that all of the stuff in between isn't executed. And it seems the goto is what is causing the problem. Of course there are a number of ways to handle this, is my read of the situation correct?

I wonder partly because the AssertNoOALError function isn't mentioned as being problematic and it does the same thing.

I only have one device and really can't take it "out of service" to test 3.0, so I'm a bit stuck helping confirm this on device. Sorry.

[Delirium]

Originally Posted by tawpie

And it seems the goto is what is causing the problem. Of course there are a number of ways to handle this, is my read of the situation correct?

Yes, you are correct. The good news is SoundEngine will work, at least in the Simulator, with a few minor modifications, which should allow many legacy 2.0 apps to run under 3.0 if one chooses to compile for it. I'm betting it works on the device as well. Like you, I have no way to test it yet though.

[Smittens]

I can now confirm that this fix lets it work on 3.0 devices as well

[Delirium]

Originally Posted by Smittens

I can now confirm that this fix lets it work on 3.0 devices as well

Excellent! Thanks for trying it out!

[gizmotoy]

I see the SoundEngine stuff has been taken care of, but has anyone tried the OpenGL portion of Crash Landing on 3.0, such as the initialization stuff and Texture2D? I'm still timid about updating my personal phone (and development device) to a beta OS. With Apple's recent announcement, though, I may have to.

As an aside, if someone with 3.0 is willing to try out my game to see if it works, I'll provide a promo code. If it's broken, then I'll have to update to the Beta ASAP.

[Mich]

Thank god this thread exists =)

Thank you very much for posting a fix.

I took the changes posted here and made a small tweak. Instead of replacing all the AssertNoErrors in the code, I changed the macro:

#define AssertNoError(inMessage, inHandler) \
if(result != noErr) \
{ \
printf("%s: %d\n", inMessage, (int)result); \
}

Compiling fine and dandy now!

[clarexine]

Hello.

I think the right way to correct the code is to move some declarations.

line 687 : OSStatus err = AudioFil....

becomes

line 687 : err = AudioFil....

Move the declaration "OSStatus err" after line 669.

Do the same with line 723.

[RegularKid]

Ok, I'm slightly confused

So, Apple says that you must test your app on 3.0 to ensure no compatibility issues are present before submitting now. Well, if we have to change code in order to fix problems for 3.0, how does one go about testing and submitting a 2.2.1 app?

Here's what I'm thinking: So, I've got a 2.2.1 app using OpenGL and SoundEngine ready to submit, but now I've got to download and test on 3.0. Obviously compiling and running the app works fine XCode 2.2.1. But now code needs to change when testing in XCode 3.0? Since it looks like you can't submit an app built with XCode 3.0, what's the point of testing it on there? Wouldn't changing the code break being able to build in XCode 2.2.1?

I've submitted apps before using 2.2.1 and the current app uses the same base code as those ( OpenGL and SoundEngine....no Interface Builder stuff ). Can I not assume that my new app will work fine on 3.0 and just skip the test and submit as is? What's the chance that it won't work on 3.0? It should be fine given the other apps using the same code base already work, right?

Anyone have any thoughts on this?

Thanks!

[wuf810]

Originally Posted by RegularKid

Ok, I'm slightly confused

So, Apple says that you must test your app on 3.0 to ensure no compatibility issues are present before submitting now. Well, if we have to change code in order to fix problems for 3.0, how does one go about testing and submitting a 2.2.1 app?

Here's what I'm thinking: So, I've got a 2.2.1 app using OpenGL and SoundEngine ready to submit, but now I've got to download and test on 3.0. Obviously compiling and running the app works fine XCode 2.2.1. But now code needs to change when testing in XCode 3.0? Since it looks like you can't submit an app built with XCode 3.0, what's the point of testing it on there? Wouldn't changing the code break being able to build in XCode 2.2.1?

I've submitted apps before using 2.2.1 and the current app uses the same base code as those ( OpenGL and SoundEngine....no Interface Builder stuff ). Can I not assume that my new app will work fine on 3.0 and just skip the test and submit as is? What's the chance that it won't work on 3.0? It should be fine given the other apps using the same code base already work, right?

Anyone have any thoughts on this?

Thanks!

running on 3.0 is not the same as compiled for 3.0. Therefore you compile your app using SDK 3.0 but build for 2.2. If it works then its fine to "run" on a beta 3 device and anything below 3.0...

M.

[tawpie]

I think the whole point of early testing with 3.0 on a device is so that there isn't as large a pool of broken apps when 3.0 releases to the general public and so that we devs don't go as ballistic when it takes forever to get an update through after the firmware does go live. I would plan on the queue being massive and the wait times long.

FWIW, I finally took a bunch of deep breaths and upgraded my only device to 3.0b5 after I discovered that my app was broken on the simulator. Phew, it's fine on the device. But I had to test because I send touches through to a scrollView, something that is explicitly mentioned in the release notes. Along the way I discovered two methods that are depreciated so I'll take care of them now. SoundEngine with fixes, BTW, works fine on my 2G iPhone. THANKS and props to Delirium.

Does anyone know if there is a thread started anywhere that lists whether or not a particular app actually breaks with 3.0b5 on device? I checked out the apps I own, and several (Ocharina being the most notable) are broken. Smule knows about it already. Anyway, it might be nice if those of us that have upgraded our devices and own apps were to give a warning to those devs that have broken apps but maybe like me are afraid of bricking their only test platform.

[tawpie]

@Smittens: I just happened to stumble across this on my gen1 iPhone. Let it go to sleep, or force it to sleep with the top button. Can you check if sounds still work after wake-from-sleep? Grazie

In my app the sound no longer sounds at all after wake-from-sleep.

Anybody else observe this with soundengine?

[tawpie]

UNCLE!

I had originally used soundengine because it provided "low latency" sound which allowed me to sync it well with my user interface elements.

But the sysSounds example works well enough for my rather elementary needs so I gave up and switched to system sounds for OS 3.0.

[minisamus]

Originally Posted by tawpie

@Smittens: I just happened to stumble across this on my gen1 iPhone. Let it go to sleep, or force it to sleep with the top button. Can you check if sounds still work after wake-from-sleep? Grazie

In my app the sound no longer sounds at all after wake-from-sleep.

Anybody else observe this with soundengine?

I can confirm I am seeing this with 3b5, I don't remember seeing it on 2.2.

[tawpie]

Originally Posted by minisamus

I can confirm I am seeing this with 3b5, I don't remember seeing it on 2.2.

Just an update... I installed 3.0 GM and behold, sound works after wake from sleep. SoundEngine sound that is, I'm testing with my "old" binary. This is a launch day iPhone.

Sway Lite's selection screen still has the hickup in scrolling the selection islands while the background music is playing. But magically at least it'll let the game play.

Might suggest retesting with 3.0 GM before you tear too much hair out.

[DenVog]

Originally Posted by Mich

I took the changes posted here and made a small tweak. Instead of replacing all the AssertNoErrors in the code, I changed the macro:

#define AssertNoError(inMessage, inHandler) \
if(result != noErr) \
{ \
printf("%s: %d\n", inMessage, (int)result); \
}

Compiling fine and dandy now!

Thanks for sharing this. HUGE help.

[GabeJacobs]

Originally Posted by DenVog

Thanks for sharing this. HUGE help.

genius

[jayaprada]

It Works For me on device also...

[Alexman]

strange i have cpu_capabilities.h error

have tried all fixes