LWN: Comments on "First Ubuntu Touch Tablet Brings Convergence at Last (Linux.com)"

First Ubuntu Touch Tablet Brings Convergence at Last (Linux.com)

JanC_ — Sun, 14 Feb 2016 16:58:04 +0000

They do also support the 2013 Nexus 7, which has a Snapdragon SoC, if you prefer that.

First Ubuntu Touch Tablet Brings Convergence at Last (Linux.com)

dlang — Wed, 10 Feb 2016 22:40:30 +0000

check what keyboard type you have set in android. If you set it to the right hardware, you just hit the numlock key on the keyboard.

If it's not set to the right keyboard type (IIRC, it's something like hardware keyboard), it doesn't recognize numlock, cursor keys, etc.

First Ubuntu Touch Tablet Brings Convergence at Last (Linux.com)

dlang — Wed, 10 Feb 2016 19:32:05 +0000

Many apps would be very happy running in a VM (probably possible by installing the android SDK and using it, but it really should be made into a trivial encapsulation that lets you just install things from an app store)

There are a few that really need deep integration into the rest of the system (VPN apps for example), but they are very much the exception.

There are a lot more that need access to files, but since Android supports removable SD cards, this should be trivial to provide.

First Ubuntu Touch Tablet Brings Convergence at Last (Linux.com)

flussence — Wed, 10 Feb 2016 15:14:05 +0000

I tried using a Bluetooth keyboard on Android some time in the past, but it had one flaw that rendered it unusable: numlock is on by default, with no way to turn it off.

The keyboard I have is one of those laptop-style overlapped-layout-to-save-space pains, so half the letter keys were unreachable.

First Ubuntu Touch Tablet Brings Convergence at Last (Linux.com)

paulj — Wed, 10 Feb 2016 13:36:42 +0000

Awesome. Great that it's being worked on.

First Ubuntu Touch Tablet Brings Convergence at Last (Linux.com)

halla — Wed, 10 Feb 2016 11:35:01 +0000

There's still a lot of work to be done, and it's still not simple, at all. But progress is being made: https://github.com/shashlik

First Ubuntu Touch Tablet Brings Convergence at Last (Linux.com)

paulj — Wed, 10 Feb 2016 10:52:32 +0000

+1. I really don't understand why I can't trivially install an Android runtime as easily as I can a Java JRE on any regular Linux distro, and run Android apps. It's just Java, mostly, right?

A couple of years ago when I asked that, people pointed at the GOOG specific kernel interfaces Android userspace required (binder), but my understanding is that the absolutely-required ones are now in mainline and that Android userspace should run on mainline kernels? So... what's the obstacle now?

First Ubuntu Touch Tablet Brings Convergence at Last (Linux.com)

dlang — Wed, 10 Feb 2016 01:08:24 +0000

it's incredibly annoying that vendors will make an Android app, but refuse to consider ever supporting Linux.

We really need the ability to easily run Android apps in Linux.

First Ubuntu Touch Tablet Brings Convergence at Last (Linux.com)

drag — Tue, 09 Feb 2016 16:26:35 +0000

That is correct. I've used ConnectBot, which is very nice. Android's support for bluetooth and various choices in onscreen keyboards provide a decent experience.

Termux is very convenient to use and will cover most of what you need from 'posix' environment. If that isn't good enough then I've had Debian chroot environment for Android since the very early days. Since then I've used full fledged Arch installs on Android with very good results. My last Arch install consisted of a disk image I built on my regular PC and copied over. A simple shell script would take care of mounting it and setting up the environment and executing the chroot command as the final step. I could then ssh into it or access it directly on the console.

The only really weird things you are likely to run into is that part of Android sandbox they split applications up into individual user accounts. Network access is based on group membership, so if you are running your applications out of chroot as a 'regular linux user' then you'll have to make sure it's in the correct groups.

The reason why you have to 'root' your phone is often (ignoring the nefarious 'phone companies are our enemies' reasons for now) because giving these sorts of rights to users are counter-productive. They don't care about root, they don't want root, and having access like that just opens up security holes and problems for users that would have no benefit from it.

The ideal solution, for 'traditional' Linux fans, though would be to integrate something like 'LXC' natively into Android so that you could have the full fledged network stack and other things just for your 'distro install'. This way you could combine it with a VPN tunnel and gateway so that you can have a persistent network presence for that container that is independent from however you connect to the internet, be it wifi, usb-net, LTE, or whatever.

This seems to me a much more interesting problem to solve then trying to convince my SO to depend on a beta-preview OS that has, essentially, been under development for over a decade now.

As far as 'loss of functionality' goes:

My latest Android adventure involved purchasing a cheap ELM327 ODB2 to Bluetooth adapters for troubleshooting a check engine light on my friend's car. With this and a 96 dollar Android Tablet I know I can have it setup and getting realtime results from the car's sensors with about 5-10 minutes of work. This work involves setting up the software (install it from playstore) and finding the ODB2 port on her car to plug into.

With traditional Linux distro there is a proprietary application I can install that _may_ be compatible with my Fedora laptop, but it is extremely unlikely it is going to work without some serious massaging. My choice would be to install one of the pyodbd and use a simple gui or write a simple one myself. I still don't benefit from the databases of ODB2 sensor codes, but this is not a huge deal. It just is going to be inconvenient. Needless to say whatever I want to use is not going to be present in any distro's repositories. So this means I will be installing what I need from tarballs and code repos. I still will have to probably tackle hacking on the bluetooth/dbus stuff a bit to make sure that Linux recognizes the bluetooth device correctly as well.

Of course I will still hack on my Linux laptop to get everything working well. The idea of being able to do data collection over a long period of time from my car is interesting to me. But for the purposes of my immediate goals of 'getting my friend's car working' it would of been a major fail.

That's the problem with the 'long tail'. There are relatively few people, percentage-wise (compared, to say, need to connect with a Exchange server), that still wants to know how cars work and are savvy enough to 'DIY' repairs correctly. However when you deal with Android's numbers... you are looking at nearly 2 billion Linux devices sold. At that scale the 'long tail' of "how software can solve individual people's problems" gets rather fat. Very fat. All of a sudden you have a massive market for every trivial 'nitch'. And since software is handled in a very distributed fashion with compiling, testing, and packaging done by 'upstream' and the play store just signs them and compiles them into one service, then you are able to support a many more times the amount of applications that a Linux distro can possibly provide.

Sailfish OS is probably very nice and all that and having Android compatibility is a VERY smart move and is laudable... however you could of provided the same environment and same functionality by using Android as the base and installing Sailfish OS on top of it and it would of made it much easier to be enjoyed by a larger number of people. Android is still Linux and it is still a FOSS operating system for the most part.

I am not trying to dissuade anybody here from hacking on the OS they want, but merely trying to help put things into perspective.

First Ubuntu Touch Tablet Brings Convergence at Last (Linux.com)

pwsan — Tue, 09 Feb 2016 09:13:10 +0000

No one forced Canonical to support an SoC from a vendor that's unfriendly to the free software community. Canonical could have spent their resources supporting a tablet using an SoC with better upstream kernel support, developer boards available and at least some public documentation, as some of the Qualcomm Snapdragon chips have. Indeed, with the lip service that Canonical puts on their site about free software and community developers, one might have expected such a policy. Too bad; an opportunity is lost, and the point that companies don't really care about good upstream kernel support is amply proven to SoC vendors.

First Ubuntu Touch Tablet Brings Convergence at Last (Linux.com)

ay — Mon, 08 Feb 2016 21:11:26 +0000

That's not really how consumer devices like tablets are developed (at least outside of Apple). BQ (not Canonical) developed this thing, they likely did it by leveraging an ODM's design (as usual) since almost no one designs tablets/phones in-house. That ODM already chose the SoC and other components, had a complete representative system running Android, etc. and most likely all BQ did is put some branding on it and handed it to Canonical to develop on. Almost all ODM designs are Qualcom or MediaTek-based (none of the other SoCs are competitive in this market) so none of this is particularly surprising.

First Ubuntu Touch Tablet Brings Convergence at Last (Linux.com)

b7j0c — Mon, 08 Feb 2016 19:48:06 +0000

terminals for android have been around forever. try ConnectBot.

you can even install a kinda-chroot with Termux, which even has its own package manager with a bunch of upstream stuff built for android (arm)

i can connect an external keyboard to any of my android devices. i own an adapter that allows me to plug any usb keyboard into the microusb port, which even has a passthrough for the other side of the usb cord, so i can continue to power everything. using the app EHK Pro, i can hotplug any keyboard i own, thus creating a nearly-full dev environment

at this point, all of the above is suerly far better supported on android than elsewhere

First Ubuntu Touch Tablet Brings Convergence at Last (Linux.com)

sbergman27 — Sun, 07 Feb 2016 00:51:20 +0000

Honestly, it was the Free drivers which drove me back to manufacturer firmware after 1 year of using the Open Source-Cyanogenmod firmware. Having Free drivers is one thing. That was not a problem on my Note 2. Having Free drivers that work well is another. Be careful what you wish for.

First Ubuntu Touch Tablet Brings Convergence at Last (Linux.com)

pwsan — Sun, 07 Feb 2016 00:28:02 +0000

Canonical could have selected an SoC with available community development boards and public documentation Instead they chose MediaTek. Not encouraging for a company that claims to "believe in the power of open source software" and which "could not exist without its worldwide community of voluntary developers."

First Ubuntu Touch Tablet Brings Convergence at Last (Linux.com)

anselm — Fri, 05 Feb 2016 23:57:20 +0000

Only the most hardcore 'traditional Linux' users are willing to tolerate the loss of usability you get from having a 'linux distro on your phone/tablet' style approach. Android is already mature and does a much better job.

My phone runs Sailfish OS, which is much closer to a traditional Linux system than Android is. For example, it is straightforward to ssh into the phone over USB and poke about using a shell (bash, not busybox); no “rooting” required.

Sailfish OS runs many Android apps quite well, so any “loss of usability” as compared to Android is tolerable if not virtually non-existent. My SO, who is by no means a “hardcore traditional Linux user”, is quite happy with her Sailfish-based phone; the main app, outside of what came with the phone, that she uses is Android-based WhatsApp, and that seems to work with no hassle (I don't use it, myself).

First Ubuntu Touch Tablet Brings Convergence at Last (Linux.com)

drag — Fri, 05 Feb 2016 18:10:50 +0000

> I would be amused if it would be possible to build a free system on current
embedded hardware.

Is this not good enough for you?

http://linux-sunxi.org/Olimex_A20-OLinuXino-Lime2

I know that 'allwinner' is not the best company out there from a GPL compliance perspective, but it's cheap and it's supported by open source software to a great extent.

That way you can avoid the new 'mini backdoor operating systems' that AMD and Intel are now embedding in their consumer products.

> Canonical is correct with driving/supporting convergence - maybe
a more classic interface with new input support would suite my
workflow better - but I hope to see more such devices.

The chances of Linux devices that are not Android making any significant inroads (ie: anything remotely approaching 1% of any market) is vanishingly small. Only the most hardcore 'traditional Linux' users are willing to tolerate the loss of usability you get from having a 'linux distro on your phone/tablet' style approach. Android is already mature and does a much better job.

Luckily Android is a FOSS Linux operating system and using that as a base then it's relatively easy to tack-on 'a more traditional interface' on top of it's normal UI. For to be 'blob free' you still need to have open hardware and replace the proprietary 'GAPPs' that provide some application APIs that tie your phone into Google services. Luckily much of those APIs are already replaceable with free software and 'more open' hardware is available. It's a lot easier to solve those problems then it is to try to write your own competitive OS from scratch, convincing people to buy-in, and THEN solve the problem of APIS and THEN solve the problem of hardware blobs.

If Canonical can use their influence to make the hardware for their devices 'more open' then that certainly deserves praise and our money, though. It certainly will be nice if they succeed.

This is one of those situations were we 'need to be change you want to see'. This means spending money and time on devices that respect freedom, rather then whatever is the easiest, cheapest, and provides the most ideal performance.

First Ubuntu Touch Tablet Brings Convergence at Last (Linux.com)

JanC_ — Fri, 05 Feb 2016 13:55:20 +0000

What (existing) hardware are you thinking about?

All the phone/tablet hardware that is (semi-)officially running Ubuntu right now is existing hardware originally designed for Android (and maybe some x86 tablets if they work with Android and/or open source drivers?).

First Ubuntu Touch Tablet Brings Convergence at Last (Linux.com)

JMB — Fri, 05 Feb 2016 12:31:36 +0000

If anyone can provide information about the general situation that may
be interesting.

When looking at the desktop with the use of binary blobs (maybe due
to contracts/patents) in all `free' current graphics driver (even Intel's,
if I am not wrong) and the problematic BIOS situation (UEFI, secure boot),
I would be amused if it would be possible to build a free system on current
embedded hardware.

Canonical is correct with driving/supporting convergence - maybe
a more classic interface with new input support would suite my
workflow better - but I hope to see more such devices.

If they would ever reach a market share above 10%, maybe they are in
the position to make systems more free due to their influence - but
otherwise they are not to be blamed if the question of free driver support
would stay a joke.

Our world is not yet perfect - it's up to us all to change it. ;-)

First Ubuntu Touch Tablet Brings Convergence at Last (Linux.com)

ssam — Fri, 05 Feb 2016 12:30:12 +0000

Not so promising:
https://git.kernel.org/cgit/linux/kernel/git/torvalds/lin...

But this looks good:
https://git.kernel.org/cgit/linux/kernel/git/torvalds/lin...

First Ubuntu Touch Tablet Brings Convergence at Last (Linux.com)

Karellen — Fri, 05 Feb 2016 12:06:52 +0000

I read the article, and had a bit of a look around the net for more info, but couldn't figure it out for myself, so... does anyone know if the hardware is fully supported with Free drivers in the upstream kernel?

...OK, stop laughing now.

Does anyone know how hard Canonical *tried* to push to use hardware with Free drivers? Did they try at all? Or am I still waaay too naïve by having any hope whatsoever in this regard?