The mobile app ecosystem has outgrown it's original purpose to run software in a constrained env. Phones today are more powerful than my engineering laptop in university 15 years ago. The app ecosystem appeal today is reach, platform lock-in, and great APIs.
For example. I _want_ to run Linux phones even without all the apps & convenience, except Signal messenger. I am unable to use Signal without first registering through a mobile app. I suspect the desktop version will run fine-ish (proton after all). But at the end of the day, adoption will increase if mobile apps had a compatible desktop version on a Linux phone.
It's possible to register a Signal account using signal-cli. Available on Flathub, and also Alpine repos. You can use a mobile number or a landline. I use Signal Desktop on a Librem 5 and everything works except the camera (for video calls) unless I plug in a webcam. The camera does work with other apps though.
I have made people mad by saying it, but it remains true: Every developer hour wasted on an Android ROM is an hour not invested in a platform free of Google's control.
Google likes Android ROMs because they pacify the developer community from working on real competitors, while not presenting any meaningful threat to their control of the majority of Android devices. The MADA that prevented OEMs from shipping AOSP is probably dead but what hardware manufacturer is going to risk Google's ire by shipping something.
My Android phone prevents me from taking screenshots if an app author doesn't want me to.
My Android phone prevents me from recording phone calls at the request of my carrier, even though it's totally legal for me to do so in my jurisdiction.
> My Android phone prevents me from recording phone calls at the request of my carrier, even though it's totally legal for me to do so in my jurisdiction.
my GrapheneOS phone has the record button. :) though I have to obtain all-party consent to do so legally in my jurisdiction (but that's my responsibility.)
I also strongly felt this when support for sideloading apps got dropped, and from my personal experience of dealing with rooting and working around play integrity. It shouldn't have to be like this.
Agreed. So get to it and design/built some worthwile ones.
EDIT: That was obviously not an order to the the parent, but more a lamentation about and call to the industry. Sorry kids; I sometimes forget that the binars are allergic to ambiguities. :)
The only thing that keeps me on Pixel is Google's astrophotography mode. Put the same quality camera (app and hardware), and I'm there. I'll get there faster if there is an Ektachrome and Tri-X film emulation setting. I miss the colors of film, but do not miss the chemistry or expense.
Yes, the steam deck has ignited the usecase for the portable linux machine for the normal user. Now we just need great linux on arm support and then I can run a version claude code on a portable arm device and have it control my whole device for me all day. I hope this happens sometime soon!!!
Maybe I'm misunderstanding this, but what would be the advantage of running straight Linux versus an AOSP-based mobile OS? Like, why not just keep the great apps that do run on there and ignore the Play Integrity ones that don't. Does it have to do mainly with just the governance of AOSP (i.e. Google)?
Only issue is it’s so hard to use a Linux phone as a daily driver. I have a librem 5, but I admit it’s just too raw of an experience for me to use as a daily driver.
(Warning: Am only a software/product engineer, playing dilettante here, not an actual marketing/business expert.)
Awhile back, I was thinking that one pragmatic way to get this viable Linux smartphone moving might be for hobbyists to focus on getting one easily available, affordable device working fully with pure Debian or PostmarketOS (no closed drivers or other modules, and preferably no blobs) and with Purism's Phosh.
Then that would boost contributions to, and demand for, Purism's open source platform/components for Librem 5 (and whatever the successor hardware would be).
If the cheap hardware is something like PinePhone, I'm just going to handwave that maybe this device won't cannibalize much sales of Purism's premium devices, but instead the community investment into the platform will effectively generate much higher net demand for Purism's premium products. With higher volume, Purism could maybe also hit more accessible price points.
If the Purism hardware demand happens, then there may be competing hardware entrants. And they will have to compete partly on being trustworthy and aligned with the interests of the kinds of customer who want to run a non-Apple, non-Google device. Where Purism should have a head start in credibility and goodwill. The new entrants will have to contribute engineer time (possibly: pay community contractors) to getting their device to work well with this platform, and be expected to upstream all of it as open source to the platform mainline, if they want to be attractive to these customers.
(I'm not saying the cheap device has to be PinePhone; that just seemed the most likely one at the time. It could even be something like an older popular Pixel model, with many unlockable-bootloader units available cheap on eBay, for which people are able to assemble/develop open source drivers. Or maybe GrapheneOS will get their own device built, and it can also be used for this non-Android-based open Linux platform.)
I fully think an amazing consumer-targetting device could take over like a storm if done well, if ambitiously done, with an aggressive software stack.
But. I think what we should ask for now should be simpler. Let this be an alpha geek toy, let folks fiddle with some basic devices boards that can do the thing. The work on PinePhone, Mobian, others is good pioneering work, alas largely held back by there just being so few decent devices for folks to play with. The driver situation keeps making hope here impossible.
It's not a high hope, but Qualcomm has a QCM6490 chip is maybe a rare hope. A chip that is somewhat buyable by regular makers, an extended life version of the Snapdragon 778G. It's pretty modern, and comes with very featureful connectivity hardware. We're seeing variants like non-cellular Radxa Dragon Q6A in the field. Particle has a new Tachyon board you can buy with it. https://www.cnx-software.com/2024/07/31/tachyon-business-car...
It's just stunningly rare alas that folks can make systems with vaguely modern cellular chips. The cores are just not available generally. Sure it's be great to have a well produced Linux phone that is super consumer acceptable with a great OS build out, a new or revived Maemo or a Jolla Sailfish: folks who can go sign the NDAs and make a consumer device but have it be Linux. But I think for this dream to really take hold, humanity needs to be afforded some possibility to have an honest shake, some chance to be a little closer to the machine than typical cellphone bargain. The lack of cellular chip availability has been so so damning to this quest. And here is one counter-example, a crack in the wall, where I see flowers and hope grow.
There was some real nice moments where it seemed like maybe some Snapdragon cellphones in general we're getting Linux support to some level, in mainline, just for the base stuff. No cellular. Unclear to me but it seems like maybe those were just the very barest of beginnings; whether any peripherals at all work or whether there was even a screen is unclear. The trickle of releases also seems to have died off. FWIW though, I will note the previous Fairphone 5 does use the above QCM6490. https://www.phoronix.com/news/Linux-6.1-Arm-Hardware
>Banking requiring an Android or iOS Device for 2fa
>My local postal service requiring an Android or iOS Device
to unlock those postal delivery boxes
>My local public transport requiring a Android or iOS Wallet app for my ticket to be used
>My Health Insurance Provider requiring an Android or iOS App to see my own insurance data
This is my daily struggle. All of these companies refuse to engage with you on this topic, you get a canned response from support that's it. How do we even win this fight? As far as I can tell we've already lost.
Total agreement with the article's conclusions. I'm an Android developer who once had about six apps listed in the Play Store. But as time passed, maintenance became more and more baroque, and a simple Android version change required me to rewrite all my apps or lose my listings. Like many developers, I gave up.
Then Google announced a decision to disallow sideloading (not clear when this will take effect) and many tablet/cellphone manufacturers intend to disallow bootloader unlocking. If all this happens, it basically closes the Android platform to anything but "official" software releases.
Consider this from my perspective. My first computer was an Apple II in the late 1970s. I could do anything I wanted with it, and I did. But over the decades I've watched the world of software development -- with the exception of personally owned Linux machines -- gradually turn into a walled garden.
What can I say -- it sucks the joy out of programming.
Where are the open source planes, trains, and automobiles? Medical equipment? Nuclear reactors? Open source cannot afford the quantity control/verification need for these domains. It’s the same for phones. At best you’re going to get an insecure mess.
Apologies if the idea is absurd, but wouldn't a Linux handheld without a cell modem be easier to build and distribute? Think something of an analogue to iPod Touches, which were iPhones sans the the phone part.
This would skip a lot of the regulatory red tape, bring down costs, and make the devices more accessible so they’re in more developers’ hands. They’d have to tether from your primary phone which isn’t ideal, but workable.
Pardon my potentially naive question, but would Samsung ever develop their own OS? I imagine they're not necessarily happy about some of the latest changes to android.
Mao said "Let a thousand flowers bloom, a hundred schools of thought contend".
Then he killed off all those naive ones who stepped out.
This is more or less the capitalist/liberalist/colonial/MAGA model from time immemorial: preach "freedom" to put yourself in a indispensable place.
Then impose fascism with long-suspected hierarchies.
As someone willing to put up with all manner of nonsense (overpriced/underpowered hardware, clunky UI, endless troubleshooting), battery life on mobile Linux devices alone prevents me from using them in the real world.
Is there a single Linux phone/tablet that can last an 8 hour day of actual use? Librem/Pinephone/Juno can't. My uConsole can't. Different category, but my MNT mini laptop lasts like 4 hours and can't be left in standby for too long or it drains to zero.
Meanwhile, it's been 10+ years since I've worried about daily battery life on mainstream mobile devices, even my 3-5 year old ones. I can fall asleep with Youtube playing and it's still playing when I wake up. I'm certainly not here to dunk on Linux phones. I want one! But if someone willing to put forth above average effort to use these devices can't realistically daily drive them, who can?
With my, atypical maybe?, use, I get up to 2 days on a 4KmA battery (Gigaset GS5, SailfishOs). Sometimes I'm down to 1 day if I do social media scrolling.
Hi. I have a Google Pixel 3a running PostmarketOS https://postmarketos.org/ and it holds up pretty well. My phone lost 20% today with light usage and will maybe lose 60% if I scroll social media a lot.
I was actually surprised it is this good. I reinstalled recently and before the reinstall I had much worse battery life (Maybe 8 hours with normal usage). I think it was because of Syncthing running in the background.
It is also possible to use s2idle suspend which will improve the battery life even more but you will not be able to receive calls during suspend (though that may also be fixed in the future)
Here is an idea I thought long and hard about for the last 3 seconds....
Say one, rather than making the entire phone modular, adds just one cartridge slot. Have it span the bottom half of the back of the phone and be a few mm deep. Cartridges can have 4 form factors. 1) flush with the back of the phone. 2) stick out from the back. 3) increase thickness of the entire phone. Or 4) like 3 but comes with the same slot as the phone so that one can stack cartridges.
The first base phone should be functional by it self but have really low specs. A slow cpu, little memory, little storage, small battery. It may even run on android and have a ton of preloaded apps no one wants. Ideally the most expensive component should be the cartridge connector.
And then, here it comes, you've already guessed it! The entire linux computer goes on the cartridge.
Have a similar dock that turns the cartridge into a desktop computer and a dock that connects it to your PC.
Software development would be glorious.
In the initial demo it should run Windows! This will send a strong signal to other otherwise uninterested parties that this is a real computer... finally...
While official builds should probably exist let other vendors go wild building their own proprietary closed source cartridges.
There should be infinite possibilities. People will make things we cant imagine. Stuff we will never see on flagship phones because 99% doesn't need it.
Some might simply but badly want usb ports.
Stupid example: I have a digital camera, I have to plug it into a computer and do all kinds of things before they may appear on my server, like booting the machine, opening apps and figuring out where the hell folders are. The pictures are great but not that much better than my phone which can conveniently send them places. But what I really need is to just plug in the camera and have the technology figure out which are the new images and upload them. It should require zero screen time.
The next guy might want an ethernet port, hdmi, serial, scan barcodes by pressing a real button that also unlocks and opens the correct app. You might even have a bulky cartridge that prints receipts. A large antenna and/or a week worth of battery. I'm not at all sure if people want it but a cassette player would be possible. A boom box with atx drive bays. etc etc
Then when you buy the next generation or are bored playing with it, the screen is cracked and the battery is worn out you turn it into a security camera that works when the power is cut and can send [picture] sms, make phone calls and play threatening messages to intruders.
92 comments
[ 3.7 ms ] story [ 96.6 ms ] threadFor example. I _want_ to run Linux phones even without all the apps & convenience, except Signal messenger. I am unable to use Signal without first registering through a mobile app. I suspect the desktop version will run fine-ish (proton after all). But at the end of the day, adoption will increase if mobile apps had a compatible desktop version on a Linux phone.
AOSP is open and is a much better starting place than anything else.
Google likes Android ROMs because they pacify the developer community from working on real competitors, while not presenting any meaningful threat to their control of the majority of Android devices. The MADA that prevented OEMs from shipping AOSP is probably dead but what hardware manufacturer is going to risk Google's ire by shipping something.
My Android phone prevents me from recording phone calls at the request of my carrier, even though it's totally legal for me to do so in my jurisdiction.
I'm not loving where this is all going.
It's worse. An app author can even be notified if a screenshot was taken.
I'm assuming (hoping) it's Android letting me know and not the app passive aggressively side eyeing me.
I told them that their app prevents this. To their surprise.
I told them that I would use the web site and they were happy that there is a workaround for their own limitation.
I had other wild stories with this otherwise good bank.
my GrapheneOS phone has the record button. :) though I have to obtain all-party consent to do so legally in my jurisdiction (but that's my responsibility.)
Then again, this is also what makes me almost throw my Android phone against the wall when I try to do the same on that phone.
(Linked from the post: https://forum.syncthing.net/t/discontinuing-syncthing-androi...)
Agreed. So get to it and design/built some worthwile ones.
EDIT: That was obviously not an order to the the parent, but more a lamentation about and call to the industry. Sorry kids; I sometimes forget that the binars are allergic to ambiguities. :)
Rock solid. Every few year feature updates, only security fixes otherwise.
Awhile back, I was thinking that one pragmatic way to get this viable Linux smartphone moving might be for hobbyists to focus on getting one easily available, affordable device working fully with pure Debian or PostmarketOS (no closed drivers or other modules, and preferably no blobs) and with Purism's Phosh.
Then that would boost contributions to, and demand for, Purism's open source platform/components for Librem 5 (and whatever the successor hardware would be).
If the cheap hardware is something like PinePhone, I'm just going to handwave that maybe this device won't cannibalize much sales of Purism's premium devices, but instead the community investment into the platform will effectively generate much higher net demand for Purism's premium products. With higher volume, Purism could maybe also hit more accessible price points.
If the Purism hardware demand happens, then there may be competing hardware entrants. And they will have to compete partly on being trustworthy and aligned with the interests of the kinds of customer who want to run a non-Apple, non-Google device. Where Purism should have a head start in credibility and goodwill. The new entrants will have to contribute engineer time (possibly: pay community contractors) to getting their device to work well with this platform, and be expected to upstream all of it as open source to the platform mainline, if they want to be attractive to these customers.
(I'm not saying the cheap device has to be PinePhone; that just seemed the most likely one at the time. It could even be something like an older popular Pixel model, with many unlockable-bootloader units available cheap on eBay, for which people are able to assemble/develop open source drivers. Or maybe GrapheneOS will get their own device built, and it can also be used for this non-Android-based open Linux platform.)
But. I think what we should ask for now should be simpler. Let this be an alpha geek toy, let folks fiddle with some basic devices boards that can do the thing. The work on PinePhone, Mobian, others is good pioneering work, alas largely held back by there just being so few decent devices for folks to play with. The driver situation keeps making hope here impossible.
It's not a high hope, but Qualcomm has a QCM6490 chip is maybe a rare hope. A chip that is somewhat buyable by regular makers, an extended life version of the Snapdragon 778G. It's pretty modern, and comes with very featureful connectivity hardware. We're seeing variants like non-cellular Radxa Dragon Q6A in the field. Particle has a new Tachyon board you can buy with it. https://www.cnx-software.com/2024/07/31/tachyon-business-car...
It's just stunningly rare alas that folks can make systems with vaguely modern cellular chips. The cores are just not available generally. Sure it's be great to have a well produced Linux phone that is super consumer acceptable with a great OS build out, a new or revived Maemo or a Jolla Sailfish: folks who can go sign the NDAs and make a consumer device but have it be Linux. But I think for this dream to really take hold, humanity needs to be afforded some possibility to have an honest shake, some chance to be a little closer to the machine than typical cellphone bargain. The lack of cellular chip availability has been so so damning to this quest. And here is one counter-example, a crack in the wall, where I see flowers and hope grow.
There was some real nice moments where it seemed like maybe some Snapdragon cellphones in general we're getting Linux support to some level, in mainline, just for the base stuff. No cellular. Unclear to me but it seems like maybe those were just the very barest of beginnings; whether any peripherals at all work or whether there was even a screen is unclear. The trickle of releases also seems to have died off. FWIW though, I will note the previous Fairphone 5 does use the above QCM6490. https://www.phoronix.com/news/Linux-6.1-Arm-Hardware
>My local postal service requiring an Android or iOS Device to unlock those postal delivery boxes
>My local public transport requiring a Android or iOS Wallet app for my ticket to be used
>My Health Insurance Provider requiring an Android or iOS App to see my own insurance data
This is my daily struggle. All of these companies refuse to engage with you on this topic, you get a canned response from support that's it. How do we even win this fight? As far as I can tell we've already lost.
Then Google announced a decision to disallow sideloading (not clear when this will take effect) and many tablet/cellphone manufacturers intend to disallow bootloader unlocking. If all this happens, it basically closes the Android platform to anything but "official" software releases.
Consider this from my perspective. My first computer was an Apple II in the late 1970s. I could do anything I wanted with it, and I did. But over the decades I've watched the world of software development -- with the exception of personally owned Linux machines -- gradually turn into a walled garden.
What can I say -- it sucks the joy out of programming.
On security, your lovely propietary NT based kernels had so many CVE's and attacks that you could fill gigabytes of text with these.
This would skip a lot of the regulatory red tape, bring down costs, and make the devices more accessible so they’re in more developers’ hands. They’d have to tether from your primary phone which isn’t ideal, but workable.
This is more or less the capitalist/liberalist/colonial/MAGA model from time immemorial: preach "freedom" to put yourself in a indispensable place. Then impose fascism with long-suspected hierarchies.
Is there a single Linux phone/tablet that can last an 8 hour day of actual use? Librem/Pinephone/Juno can't. My uConsole can't. Different category, but my MNT mini laptop lasts like 4 hours and can't be left in standby for too long or it drains to zero.
Meanwhile, it's been 10+ years since I've worried about daily battery life on mainstream mobile devices, even my 3-5 year old ones. I can fall asleep with Youtube playing and it's still playing when I wake up. I'm certainly not here to dunk on Linux phones. I want one! But if someone willing to put forth above average effort to use these devices can't realistically daily drive them, who can?
I was actually surprised it is this good. I reinstalled recently and before the reinstall I had much worse battery life (Maybe 8 hours with normal usage). I think it was because of Syncthing running in the background.
It is also possible to use s2idle suspend which will improve the battery life even more but you will not be able to receive calls during suspend (though that may also be fixed in the future)
Say one, rather than making the entire phone modular, adds just one cartridge slot. Have it span the bottom half of the back of the phone and be a few mm deep. Cartridges can have 4 form factors. 1) flush with the back of the phone. 2) stick out from the back. 3) increase thickness of the entire phone. Or 4) like 3 but comes with the same slot as the phone so that one can stack cartridges.
The first base phone should be functional by it self but have really low specs. A slow cpu, little memory, little storage, small battery. It may even run on android and have a ton of preloaded apps no one wants. Ideally the most expensive component should be the cartridge connector.
And then, here it comes, you've already guessed it! The entire linux computer goes on the cartridge.
Have a similar dock that turns the cartridge into a desktop computer and a dock that connects it to your PC.
Software development would be glorious.
In the initial demo it should run Windows! This will send a strong signal to other otherwise uninterested parties that this is a real computer... finally...
While official builds should probably exist let other vendors go wild building their own proprietary closed source cartridges.
There should be infinite possibilities. People will make things we cant imagine. Stuff we will never see on flagship phones because 99% doesn't need it.
Some might simply but badly want usb ports.
Stupid example: I have a digital camera, I have to plug it into a computer and do all kinds of things before they may appear on my server, like booting the machine, opening apps and figuring out where the hell folders are. The pictures are great but not that much better than my phone which can conveniently send them places. But what I really need is to just plug in the camera and have the technology figure out which are the new images and upload them. It should require zero screen time.
The next guy might want an ethernet port, hdmi, serial, scan barcodes by pressing a real button that also unlocks and opens the correct app. You might even have a bulky cartridge that prints receipts. A large antenna and/or a week worth of battery. I'm not at all sure if people want it but a cassette player would be possible. A boom box with atx drive bays. etc etc
Then when you buy the next generation or are bored playing with it, the screen is cracked and the battery is worn out you turn it into a security camera that works when the power is cut and can send [picture] sms, make phone calls and play threatening messages to intruders.