The advantages of mobile Linux over AOSP
6 points by brn
6 points by brn
I've thought about making the jump to Linux phone for some time now, but the main challenge remains (drumroll) calls.
If you look at postmarketOS, it seems that call / sms support is very spotty. After being in an emergency situation, the phone must be able to do an emergency call. Is there a phone which can do 4G/LTE calls using a (near) mainline kernel? If not, isn't your mobile Linux really just a tablet?
Mainline... I think no, VoLTE support on mainline Linux still isn't there to my knowledge. The libhybris-based distros/phones on the other hand do tend to have pretty good calling support; so far this year I've used both an FLX1s running FuriOS and a Fairphone 5 running Ubuntu Touch as daily drivers. (Currently using the latter.)
One caveat re: emergency calls that I'd like to see addressed somehow: if your distro uses full disk encryption via LUKS (as opposed to something like home dir encryption), you will not be able to make a call of any kind until the disk is unlocked after powering on. I'd like to see some sort of minimal environment capable of making calls that you can boot into without decrypting as a failsafe, but I'm sure that would be challenging to do in a way that's reliable (and it really needs to be reliable).
Most of us keep our phones powered on most of the time anyway, but in case it's not, would be helpful to have that to fall back on.
That's fair. I haven't tried Linux mobile in a few years, but I also struggled with basic phone functionality. I'd hope things have improved over the years, though, but banks requiring iOS/Android (non-GrapheneOS) is a big blocker for me.
I used a Freerunner for about ten years, and the only problem I had with calls was after messing with some of the volume sliders in alsamixer, which made my mic inaudible until I figured out how to undo what I'd done.
I've been using a Pinephone since then. I had some nervousness when I switched from Pulseaudio to Pipewire, since the way audio would switch from the main speaker to the earpiece speaker was a bit hacky; but if anything, pipewire seems smoother.
I've been considering trying to build my own mobile device/tablety thing and have similarly found that VoLTE support is extremely minimal even (or perhaps especially?) in the world of standalone USB and PCIe LTE modems... The closest I've found is "these specific ones MIGHT work, just try them with Gnome Calls and do these random 3 things to attempt to get them to work if they don't." SMS support seems significantly better, but still ridiculously bad compared to LTE internet.
edit: and now I've done some more research and found an LTE modem that may have reasonable support (quectel eg25 - has community of linux hackers because is the modem for PinePhone)... but have been completely demoralized by finding that T-Mobile may be (tenuous leaks) sunsetting new LTE activations as early as 2028.
This article seems to mostly just handwave the development costs, security, ecosystem angle with "don't worry, it's 80% of the way there and developers are cheap". Yet we all know how much the last 20% costs...
I suspect a hostile fork of AOSP if Google did close it would happen. Even if it was i.e. just Graphene, forking Android means creating i.e. an alternative app ecosystem doesn't mean struggling to get basics like sandboxing and phone calls working; you could convince existing Android app developers to support you, you could use the work of i.e. /e/, target existing devices more feasibly than Nura's "we got it booting" level of kernel support, etc.