[Tinkerphones] New LetuxOS Kernels and some tricks and thoughts

[H. Nikolaus Schaller]

Hi Paul,

> Am 21.05.2019 um 14:12 schrieb Paul Boddie <paul at boddie.org.uk>:
> 
> On Tuesday 21. May 2019 10.22.50 H. Nikolaus Schaller wrote:
>> 
>> What else is going on in the LetuxOS eco-system?
>> 
>> We now support MIPS devices, not only ARM. Well one. The Imagination CI20
>> board. Just do "makesd ci20" to get a first SD card.
> 
> I'll have to stop procrastinating and try this now. Well done for getting to 
> the bottom of the kernel configuration problems, too!
> 
>> Behind the scenes, we have debootstrapped mipsel variants of the Letux
>> Debian images. And adapted the kernel defconfig so that we get a CI20
>> (Ingenic JZ4780) compatible kernel.
>> 
>> What is the motivation behind supporting yet another board (and not fixing
>> all others first)? The main aspect is that the jz4780 also contains an SGX
>> 544 GPU - like the OMAP5. This should help to get the SGX drivers working.
> 
> It might also help get the Letux 400 Minibook supported by a modern kernel as 
> well. :-)

Yes! This is also an important goal and comes much closer to reality now with
the CI20 up and running.

> 
> [Exciting PyraPhone and other news]
> 
>> On a very general view, we have achieved a lot, but still not enough to get
>> the LetuxOS eco-system into a self-sustaining mode. What is lacking?
>> 
>> * users are missing because software is not good enough for daily use
>> * hardware is missing because potential users complain about missing
>> high-quality software
>> * developers to polish the software are missing, because of missing (new)
>> hardware
>> 
>> You see the vicious circle? Ideas how to magically break it?
> 
> I ended up ranting a bit about this again recently. But it just takes a glance 
> at the state of the smartphone market to see where some of the problems are 
> and where we might be able to offer solutions. I'm sure most people on this 
> list already know everything that I write below, but it is sometimes worth 
> reflecting on a situation to remind ourselves of the value in the things we 
> do.
> 
> Let us start by investigating what kinds of sustainable choices are available 
> for smartphone purchasers. It might be nice to consider the conventional 
> sustainability and ethical production criteria (and we should do so), but with 
> our backgrounds we should start with an evaluation of the software on any 
> given phone because this is often the limiting factor in the phone's 
> longevity:
> 
> * Is the software up-to-date or upgradeable to bring it up to date?
> * Can the user build it all from sources?
> * Can the user deploy all the built software?
> * Will it always be possible to upgrade and use new software?
> 
> Given the way phones are produced and sold, the answer to these things is 
> generally "no". Sure, the first answer may be "yes" if you buy a brand new, 
> top-of-the-line phone, but it will eventually be "no" as well. The producers 
> just want you to replace your phone and hope that this will be an acceptable 
> solution, which it is not:
> 
> https://www.smart.uio.no/news/swap-svitsj-and-burn-baby%21.html
> 
> Also, there tend to be very firm restrictions on deploying new software, 
> particularly at the lower levels. So, it is all very well the producer 
> offering a bundle of sources on their licence compliance site, but if they 
> don't let you deploy the software, it is all a charade and fails to empower 
> the end-user.
> 
> Naturally, there are choices like Fairphone which focused on conventional 
> sustainability issues first and arrived at the software issues later. Support 
> for updated software is apparently better on the second Fairphone model, but 
> there are still fundamental limitations to that support. And currently, only 
> refurbished units are available because production has been discontinued.
> 
> With this background, it is then worth looking at "community" initiatives that 
> try and remedy the software situation. This seems to involve the following:
> 
> * Pure Free Software distributions (like Replicant)
> * Hybrid distributions that favour GNU/Linux (like PostmarketOS)
> * Hybrid distributions that favour Android (like LineageOS)
> * Application distributions derived from the Android Open Source Project
>   (like OmniROM and a bunch of others)
> 
> Maybe I conflate or confuse some of these things, particularly the latter 
> ones. But it is worth looking at which devices they tend to support. The 
> latter ones seem to have broader device support, but they also appear to be 
> focused on reworking the applications, decluttering the user experience, 
> "removing Google", and so on. Just like products with only older Linux kernel 
> support, the time will inevitably come when newer distributions and 
> applications just cannot be used.
> 
> The former initiatives are more useful because they hint at which hardware can 
> be usefully supported. In their device guides you will tend to find GTA04 
> mentioned because it actually offers desirable things for these initiatives: 
> documented and supportable hardware, software that is openly and sensibly 
> developed (within the constraints of Linux kernel development, at least). 
> Things like the GTA04 should be dream devices for these efforts because the 
> reverse engineering aspects usually required are practically eliminated.
> 
> So, this is a long way of arriving at the conclusion that the hardware and 
> software efforts around LetuxOS have natural partners in the form of software 
> initiatives targeting mainstream devices. There should also be partners 
> amongst Free Software organisations: a mailing list run by one of these was 
> the venue for my recent rants. People developing Free Software user interfaces 
> should also engage constructively with these efforts.
> 
> However, my frustration is that consumerist tendencies get the better of some 
> people, that hardware projects in particular are underappreciated because 
> people think that they can "shop around". Who cares if some project fails when 
> there is always another that might deliver? The problem with this is that all 
> of these projects will fail without dependable support from those who would 
> actually benefit from them.
> 
> So we have to get these natural partners to realise that their own efforts are 
> in jeopardy if they fail to support the initiatives being nurtured on this 
> mailing list. It just isn't a sustainable and practical solution to recommend 
> very old mainstream phones that were saved from the normal channels, maybe 
> even refurbished, as the FSF does: Free Software users need hardware products 
> to be continually available so that they may actually run their software.
> 
> And the hardware products need to be given proper attention and commitment 
> from the software projects, rather than being treated as toys to be discarded 
> when something newer comes along.

Well written analysis! I really enjoyed reading it.

LetuxOS eco system is intended to break out of this pattern by trying to develop
an eco-system of hard- and software where every piece follows a holistic view
to make things smooth and interoperable. I.e. make it easy for the user to
have something reliable. Hence my question how to make it more self-sustaining.

Unfortunately if we focus on hardware (and bootloader) we neglect kernel. If we
focus on kernel we neglect user-space. If we focus on user-space we neglect
hardware.

So this would mean that all three parts should be equally represented. Unfortunately
it is too much burden to do that all at once in parallel... There are not enough
subteams of equal strength which are cooperating on such a balanced approach.

Letux eco system has to offer:
* different hardware choices
* working kernels with good hardware support (although not perfect - e.g. camera not working well)
* different GUI options (just look at the choices of <http://download.goldelico.com/letux-debian-rootfs/?C=M;O=D>)

But as said focussing on one end defocusses on one of the others and things
start to be (bit-)rotting. The result is that even I do not use a GTA04 as my
daily phone... Because a commercial one - despite all the known issues - is
easier to use, rarely has software bugs and simply works. And usually cheaper
with better technology inside.

BR and thanks,
Nikolaus