Be Excellent To Each Other

The bug is easily explained.

Trance = breakdowns.

Simples.

Total guess and hardly my area of expertise but I'd wager this will be something to do with the handling and size of a single track/file which can be cued or reviewed (forward or r versed). Thinking about it, in order to be able to do this reasonably quickly in real time, the entire file needs to be "loaded" as it were (especially as you listen to the speeded up cue/reverse as you do it), and >90 mins is an uncommonly long play, big file. The software will be expecting a finitely long file? This limitation must come into play at some point, but in which case it's a limitation, not a bug.

On the other hand, this could well be bollocks.

Or the timestamp value only goes up to 59 minutes and 59 seconds?

I quite like trance myself, though look ridiculous playing it lol. Happy memories though; man, we partied HARD back in the distant day.

You may be on to something. I can't recall offhand how you seek within mp3 files but I can believe that you can't easily jump to arbitrary points in the file because lots of compression algorithms use information from the past to encode the current state (e.g. i-frames and b-frames in video codecs.) In other words, as you put it, the entire file has to be processed from the beginning. I can also believe the codec implementation in an old-ish car would already have been obsolete when it shipped, ie. is pretty outmoded now, so could suffer from limitations in this regard. (I'd argue it's still a bug, from the user's viewpoint, though; and that's all that really matters.)

If that's the case, it could be a buffer overflow or buffer truncation again, but when the head unit tries to seek to a point to resume playback instead.

Hmm, but on the other hand, if that was it, I wouldn't expect AE to be able to manually seek within the file, either, for the same reason... interesting.

I'd be astonished if that were true. It would imply all sorts of weird things about the programmers.

Well, maybe not that astonished.

We're not going to get much further without data. Ideally we'd need to establish some correlations, e.g.

1) played to x min y seconds: resumed fine
1) played to a min b seconds: resumed at c min d seconds

The questions are: for what maximum value of x:y does it go wrong, and what can we deduce from how a:b maps to c:d?

Too fucking right.

Shit the bed, you mean I could be right? Bloody hells bells, every dog has its day! Cheers Doc.

And we'll need an S4. You do that stuff, and I'll poodle about in it.

Interesting. Is that why it takes a while to skip to the fun bits in ['cat videos' - Ed.]?

We're in a 24/7 news cycle, nie. No time for plots and exposition, just Megyn shouting what is about to happen on a loop, forever.

Ah, Megyn. *sigh*.

More or less, yes.

In video codecs, most frames that appears on the screen is defined in terms of the frames that proceeded it (and sometimes the ones that succeed it, too.) So if you have a still shot of a talking head, most of the frames will only contain data about the mouth movements, as the rest of the frame is still. This means it's hard to just jump to an arbitrary point. Some frames reset this structure, partly to enable seeking -- you can jump to one of those fine.

If you jump to a frame that isn't one of the reset ones, the decoder might "rewind" to the last reset one then play forwards to your desired frame before resuming. This can cause slowness, especially if it's a crappy Flash applet doing the playback, as can be the case for "cat videos." Or it might just start playing the intermediate frames right away and you'll see corruption and smearing all over the screen for a second or two.

This is also why DVDs have chapter marks which you can skip to instantly -- they are basically pre-defined points the codec knows it can seek directly to and start decoding right away. But you can't navigate a DVD except via chapter jumping or ffwd/rwd, for this reason.

Disclaimer: I'm far from expert in this sort of thing, I'm glossing over a few details I know and surely a lot more I don't.

The compression is a 'track changes' thing?

Nbc, now. Or, as I like to call it "Next Bus: Communism'

Live reportage from our grotty back street. (Well, live-ish.)

Will it work? OR WON'T IT?

Yeah. It's a lot cheaper to track deltas from frame to frame than it is to store the entire contents of each new frame. This is more intuitively explained with video, which is also most of the sophisticated codec work has happened in the last 20 years anyway, but I'm pretty sure applies equally to audio.

Obligatory Tom Scott:

That's why Ghost in the Shell is only 8Kb.

Everything is 8Kb in 240p.

It is a diary moment

Real missed opportunity for a Vorsprung Durch Techno gag, there.

That would have been really neat.

YOUTUBE FAIL.

Heartly, you sound exactly like the guy who works here that everyone suspects to be a creepy rapist. Just FYI

Harsh!

I have no idea what to do with that information.

Possibly one of those things you should think in your head but go no further with.

Sorry, I should have said "potential rapist". He's under investigation.

Anyway, carry on!

That... doesn't really make it a less odd thing to say, dude.

Carry on.

Everyone is a potential rapist.

Apart from actual rapists. They're just rapists.

You sound like a radfem

I don't think they should have recast girls in the Terry Waite film.

Let's not derail the Car thread, eh chaps? Sorry for my comments.

Cars don't run on rails, you berk

Yes they do.

That still looks pretty train-y to me.

Well that didn't work.

Train car, innit?

If it helps, every time I look at the giphy I start sniggering loudly.

Lets get back on topic

Giphy "70k porsche":
http://media3.giphy.com/media/RksUScDXCA8OA/giphy-loop.mp4

That colour-change pack is extra

Useful if you have an accident; the witnesses will contradict each other.

You barstewards

Besides which, a Macan isn't a bloody Porsche.