[davidhyde]

The Bluetooth Special Interest Group (SIG) is responsible for publishing all the Bluetooth specs and you can download them here: https://www.bluetooth.com/specifications/specs/

Notice that everything appears to be there: very detailed specs and information about testing. However, when you try to implement one of these specs you quickly realize that you cannot do it with the spec alone. You need example code, base implementations, test suite software and test data to build conformant software. Unfortunately, the Bluetooth SIG hides these resources behind a membership wall. Guess what happens then? You get lots of implementations of these specs that are a little bit off and don't handle all edge cases.

If I were to wave a magic wand I would like to see Bluetooth SIG change to a donation based financial model and for them to make all resources freely available. Right now they make money from branding, certification and country club membership fees. No wonder the ecosystem is one big tire fire.

[natch]

So, you are suggesting that the Bluetooth SIG did just a fine job with the technology, no problems there, and that people like Apple don't see fit to use the money in their budget to cough up a membership fee? Seeing as how it is Apple devices on which I often encounter pairing difficulties. Super convenient for the SIG that they can point fingers like this (I wonder if that's why they set it up as you describe, ugh), but I would be skeptical of any implication that Apple doesn't pay a membership fee.

[bee_rider]

Do you have other non-Apple devices with which you encounter pairing difficulties?

My experience with Bluetooth headphones was that on an iPhone they were occasionally annoying, while on Linux they were quite poor and on Windows they were more like a practical joke. This would have been like 4-5 years ago though, maybe all the stacks have been improved.

[JLCarveth]

Honestly my linux desktop (Ubuntu 22.04) is where I run into the fewest issues regarding my bluetooth headphones. I've never had an issue connecting to them.

[ikeboy]

I've had issues where every other time I connect my Sony XM5 to Ubuntu 22.04, it decides to use HFP instead of A2DP which has significantly worse audio quality for music. I need to reset Bluetooth to fix it. None of the solutions online for disabling this "helpful" feature work.

[felurx]

This was quite an annoyance for me too. I discovered/noticed two things, but I'm on Manjaro, so YMMV: Whether A2DP or HFP is used can be controlled by changing the profile (or whatever it's called) of the audio device, in the audio settings. Take a look in the settings where you can change volume per app and all that good stuff, and try to see if there's some dropdown that lets you pick between Headset and Headphone, and all that.

Second thing: The actual default behaviour seems to be to use the profile that was used last time. But for me, whenever MS Teams is running, it decides that the correct thing is to switch it to Headset mode, because Microsoft knows what you want better than you do, and you want the option that produces shitty sound quality, but bidirectional audio.

[solardev]

Heh, that was the same issue that made me give up on desktop Linux (though with the XM3) and go back to Windows/Mac... three years ago.

It's just never going to get priority because it doesn't have the marketshare. Sony doesn't care about Linux and Canonical doesn't care about Sony.

If I were you, I'd consider an external bluetooth adapter that supports A2DP and connects to the 3.5mm plug on your computer, like this one: https://aluratek.com/universal-bluetooth-optical-audio-recei...

Look for AptX low-latency support too if you care about audio lag (for movies or games).

[bee_rider]

Hmm. The issues that come to mind are annoying pairing (Arch -- probably my fault, too manual) and some weird random disconnects on Ubuntu, but that might be some power-saving issue specific to my chip. So maybe I just got unlucky random outcomes.

[MonaroVXR]

QC35... Switch between One Plus 6 and Envy 2021 2 in 1 seamlessly.

Everything switches seamlessly from LG earbuds to Samsung Galaxy A52 5G, Nintendo Switch, Envy with Pop OS and Windows 11.

I have no idea what people are doing.

[techdragon]

The QC35 Bose headphones I have are on their second ear cups, have years of daily use and I think I have run into pairing issues maybe a dozen times. And usually it was to do with having lots of devices in my history that are actively in use and it’s understandable it might be a little confused. It’s been trivial to purge a few less frequently used devices, turn them off and on and that’s it it’s good again. They’ve been an absolute joy and honestly the only feature I’m looking for to make me consider a new set of over ear headphones is being able to listen to audio from more than one device at once… it can be paired to two devices so I can quickly pick up an incoming call, and it will play audio from one or the other depending on which is muted or plays first. It won’t switch back and forth if any background audio (even inaudible silence seems to trigger this)

[xlii]

AFAIK it's not about Apple, but about other producers either polluting spectrum or plainly ignoring specs.

There was a post about Logitech not creating a lot of interference with their MX products (https://news.ycombinator.com/item?id=31939424). Interestingly enough getting rid of Logitech devices fixed most of the BT issues I had.

[googlryas]

I have constant issues the single bluetooth device in myhouse - my Bose bluetooth QC-35s, and a google produced phone and google produced laptop which it connects to. Both of those groups should be able to get bluetooth right.

[solardev]

I can't even get my Airpods to work consistently with my iPad, or Pixel Buds to work consistently with my Pixel phone. And it gets exponentially worse if you try to connect to more than one device at a time, across vendors or even chipsets. Bluetooth is a hopeless mess.

[natch]

That’s a good point and something I could look into (very) locally in my setup. Not many non-Apple devices around except a couple of cars and one mouse but yeah they could be causing problems.

[antomeie]

SIG did not do an excellent job with the tech, that's for sure, but his point is still valid, that poor implementations of it causes further issues. Also, SIG is quite slow with new features, so companies start adding their own stuff on top of it. With Apple, things like sharing BT pairings over iCloud comes to mind, which is kind of a weird hack. They also do some non-standard switching between LE and Classic on the go.

[wvenable]

Are you sure the problem is Apple and not the device on the other side?

[natch]

Both sides are Apple, fully updated, expert user, in close proximity, good charge, blah blah blah. Maybe it’s me… or… just maybe… Bluetooth does have room for improvement.

[ricardobeat]

In my experience the W1-equipped headphones all work flawlessly, pairing is instant. The only thing that is still annoying is the difficulty of managing multiple devices & headphones, when the built-in 'right thing to do' (like connecting to the last used device) can conflict with your real intent.

[systemvoltage]

Same with USB consortium. Gotta pay like $10k to get a USB vendor ID. What the fuck were these people thinking to rally up behind this draconian standard? There is a limit to how many vendors they can have, iirc 10k.

But for all the issues with it, usb seems to work on devices without Bluetooth-like issues.

Standards should be open and free for common public and only have to pay once you reach certain company size. Things like ISO standards aren’t free.

[jfim]

10k is a negligible amount for a device manufacturer.

You'd be surprised at how expensive it is to manufacture things. A plastic injection mold to make plastic parts can be in the five figures, for example.

[blibble]

vendor id is 16-bit so should be enough for 65535 of them

[systemvoltage]

Yea a bunch of these are reserved. I remember jumping through the whole process when we were designing a usb device.

To counter my own point, it takes money and people to run a standards committee and it is important to keep in mind that funding has to come from somewhere. Here are more details of what it takes to run one: https://stackoverflow.com/a/22968431/3908009

IMO they should either be funded by gov or by large companies.

[blibble]

yes, $4k/year doesn't seem that bad for a company that's going to be producing a commercial product

(and of course if you're just hacking locally you can do whatever you want as it's an open standard)

[madacol]

Maybe in next versions they should choose 256 bits (or 160, like bitcoin) which would be the hash of a public key, an any vendor that provides a valid signature, can claim that 160 bit vendor ID. No more need for a central authority

I'm probably overengineering it, just pointing out it is technically possible

[lttlrck]

There is no obvious reason why USB couldn't have used used Ethernet MAC vendor Id, aside from politics.

Instead there could be a single registry like IANA.

[lupire]

Can you sublet a vendor id, like a taxi medallion?