> > With stalker mode comes the need for synchronization.  
> 
> Could you provide a bit more info what you mean by "stalker mode"?

It is just a term used in place of "a method of controlling the iPod to play only a single track at a time". It means 
following what the iPod is doing very closely (so that you can tell it to stop playing right at the end of the track)

> > This will leverage newer iAP commands so that we can sync quickly and 
> persistently across device reinsertions.
> 
> You should be careful with those newer iAP commands that are not supported on 
> all iPods...

Our solution would attempt to use the the best command set available. If an older iPod doesn't support certain iAP 
commands the operations would still work, just perhaps be less efficient.
 
> > One of the issues we see with stalker mode is that it overrides the iPod 
> playback engine.  So if the user were to play a list of tracks with their 
> headphones, go in the car, and then back to their headphones the user would 
> not be able to resume playing from what they were listening to before they 
> entered the car.  
> 
> This has been identified as a showstopper and we've been prohibited to specify
>  to anything that would break the "walk into car" usecase. That means any 
> metadata scanning process has to exclusivly use commands that are based on the
>  iPod Database Engine, not the iPod Playback Engine.

I think most, if not all, customers would agree with your statements (must be able to resume where the iPod left off 
previously).  

> > Because of this we plan on implementing this feature as an option where you 
> can use the iPod with either stalker mode or not.
> 
> Atleast for us, this would mean that we would never use "stalker mode" or 
> features that require it.

I think what we would have to do is allow the iPod when first plugged in to resume from the playback engine and only 
when a user overrides the playback engine (by choosing to play something new discovered through the database engine) 
would this new mode come into effect.

However, one thing this new mode does do, is "disturb" the "walk out of car" experience. If the car said you were 
playing 1/200 songs (songs picked by your software, which means a combination of songs that the iPod could never contain
 in its playback engine), and you are controlling it be resetting the playback engine between songs to enforce your 
specific selection, if the user unplugs their iPod they will be left with whatever happened to be in the iPod Playback 
engine at the time (certainly not the 200 songs that would have played when controlled by the car in stalker mode).

Regards,
Gilles

> We had tried this using our own iAP stack and some special content with a 
> easily identifiable start/end pattern. Our conclusions were that the iAP time 
> status updates reported by the iPod are not accurate enough and have too great
>  of a variance. This resulted in too many audio glitches.

The trickiest part would be if you are in fast forward mode, the iPod increases its speed as it progresses towards the 
end of the song. It would be difficult to judge when the end of the song is going to arrive.

> I also considered looking at the USB Audio stream to more accuratly identify

If you buffer the audio data slightly before playing it back it might be easier.

> the "gap" between tracks, but we didn't have the time/resources to implement 
> this... and we were concerned that these types of hacks would start to produce
>  audio gliches if Apple implements some sort of cross-fading between tracks as
>  some point in the future...

It isn't supported by Apple, so it is possible they release iPod firmware that can break this. 

> Most promising would looks to be an timer-based audio fade down, but as you 
> mentioned you may have better luck then us if you can go a few levels deeper..

Yes, that would help as well.

Regards,
Gilles

> We had tried this using our own iAP stack and some special content with a 
> easily identifiable start/end pattern. Our conclusions were that the iAP time 
> status updates reported by the iPod are not accurate enough and have too great
>  of a variance. This resulted in too many audio glitches.

Can you clarify why you need the single track playback?

Another downside is that you loss gapless playback ability (i.e. if you had an album on the iPod you do single track 
playback, there will be a much larger gap between tracks that flow into each other, I'm thinking Pink Floyd type albums)
.

Gilles