Re: creating and maintaining accessible Linux distros
Greetings to all,
From my 22 years of being a Linux user and advocate, I have seen conversations like the one's we have been having here in the last month, come up quite a bit. I have started a few of those discussions myself. They are difficult because there are lots of words used that have such a broad meaning.
Distro, a distribution of GNU/something, Linux, Herd, or maybe Freebsd, Openbsd, OpenIndiana, Darwin, you name it. The list goes on for miles. In some cases a distro is a snapshot in time of what one person thinks are the appropriate set of tools for the task they have in mind. Other times a distro is a set of recipes for constructing a system, that hopefully works at some point in time in the future.
I would say that I am a Debian user, although I have installed many hundreds of systems in the last ten years, I've only used the Debian installer once or twice. Now I have a custom netboot/virtual machine image, that I rebuild whenever I'd like. It includes the tools that I need to automate the installation of a system, in the case of netboot, or is the installed system, for VMs.
I make a point of using systems that have some form of serial console access. That might be through serial over lan, as part of the IPMI stack on many servers, JTAG, in the case of imbedded bits of kit, or lish, on the Linode platform.
As a side note, if anyone knows of another hosting platform that has character based out of band console access that isn't imbedded in a web page, I'd love to know about it.
I have stayed with Debian for 20 years because I believe the policy, and the conformance there too, produces a high quality system of software. Before that I used Slackware which required me to compile many packages myself. It was often messy and fragile. Of course, there are still certain applications that I have to compile, scripts that I need for automating tasks, and config files that I need to put in place to make the system the way I like it. Containing all these things in a framework that manages them for me is something that I would certainly like to see. This is why it is on my list to try NixOS.
All this to say, I'm not sure that the distro is the thing to focus on so much as the running system, which I would argue, the upstream packages and the distro make up. We need a lot of help in the upstream components. As Jason White recently pointed out for us, accessibility fixes in applications that are used by folks who never use accessibility features would go a very long way to making more useable systems.
I appreciate the need for running FOSS on older hardware. Experience has showed me that the best way to get hardware supported longterm is to get drivers into mainline Linux. At some point it not worth someones time to patch a driver that the other maintainers don't know how they are breaking.
I find that the real trouble with non-free software is that it is of lower quality. I am very interested in an alternative to Android/iOS on mobile devices. If that comes from Debian, Arch, Alpine, or anyone else, it doesn't matter so much how you spell the name. We need the support added to upstream projects. Debian's as good as any of the others at getting their patches accepted by upstream developers.
Samuel has done an amazing job supporting accessibility of FOSS for the last 15 years, using Debian as the platform. I know he could use whatever help people can give. I suggest we put our wait behind him, if we can get the appropriate packages brought in.
There are lots of other issues to respond to. I'm sure others will pipe up.
On Jan 5, 2022, at 4:51 AM, Jean-Philippe MENGUAL <firstname.lastname@example.org> wrote:
> tl;dr: as a Linux user since 2004, I dont think creating an accessible distributioon is a good idea. I prefer contributing to mainstream distros than creating a new one, as such maintainance may require a lot of work, the maintainers team is nearly always so small, so there is no future warranty for the end-user. While using a,n accessible Debian will be alwyas possible as there will be maintainers, using a small specific distro in the future requires the maintainer (or 2-3) to be here (and it is not certain at all one person will maintain a long time). I have seen a lot of accessible distro, maintained by a small community, few are still active AFAIK. And accessibility is necessarily maintained by a small community.
> Le 05/01/2022 à 09:04, Rich Morin a écrit :
>> tl; dr - Here are some ideas about creating and maintaining accessible Linux distros. Your mileage may vary, void where prohibited by law.
>> Some Background
>> To let folks know where I'm coming from, I'll start out with some background. I'm a sighted programmer who has been working with free software and *ix systems (mostly BSD-flavored) since the early 1980's. Although I have no particular a11y skillz, I'm quite interested in the general topic.
>> Getting to specifics, I'd like there to be a FOSS distribution that:
>> - can be installed, maintained, and used without sighted help
>> - provides both command line and graphical user interfaces
>> - runs on very inexpensive devices (e.g., old cell phones)
>> - is accessible by both blind and visually impaired users
>> - is plug-and-play and reliable enough for production use
>> - can be augmented by a wide range of FOSS applications
>> - supports "always-on" operation (like a notetaker)
>> - has a substantial developer and user community
>> - and a pony...
>> AFAIK, this collection of goals isn't met by any FOSS distributions. Some distros only work on PC or Raspberry Pi hardware. Some aren't set up to give blind users a friendly initial experience. Some have issues with driver and/or app setup and maintenance.
> debian has place to improve this, provided that persons do the work. Its first purpose is to work on most architectures, but we lack maintainers. Why not help porting packages instead of starting from scratch?
> Anyway, the purpose of "plug and play" and "inexpensive hardware" has always been the dream of persons: Debian tries do this via a modular kernel and ports mechanism, but people always want something on recent hardware, and then we experience typical hardware compatibility problems unrelated to the distro but to the FOSS in general.
>> In cases where the OS isn't specifically aimed at blind users, the a11y folks have to convince and coordinate with the boot folks to get blind-friendly startup behavior. And, if the OS _is_ specific to blind users, the developer
> Yes, debian did it. But of course, improvements stay possible, eg. creating a tasksel to install or uninstall easily accessibility features, reduce the size of the documentation so that persons find how to start installer according to their situation
> and user communities tend to be small enough that development and support are challenging.
> So imagine a still smaller community for a specific distro
>> So, I don't blame the majority of blind users for sticking with proprietary OSes. However, these don't check off all of the items on my wishlist, either. For example, neither Android nor iOS are easy to augment with FOSS software; indeed, iOS has no command line at all! Neither Microsoft Windows nor macOS can be used on a cell phone.
>> And, in any case, the needed hardware may be too expensive for many blind users.
> that is right, but here you introduce a much more global problem: FOSS and hardware, desire of people to use CLI vs capability of such hardware to run GUI, accessibility of slight GUIs (to be improved upstream).
>> OK, you get the idea, but what is the answer? Although I don't have a solid proposal to offer, I do have a speculative notion to suggest. Here goes...
>> A Notion
>> There are billions (!) of old cell phones floating around in the world. Some are staggering along on ancient Android releases; others are sitting in drawers, being discarded, etc. IMHO, this is a significant untapped resource. The processors in these phones are orders of magnitude faster than the CPUs found in the workstations of the 1980's. Better yet, they can commonly be purchased (e.g., on eBay) for $20 or so, making them affordable by a lot of blind folks around the world.
> But are you talking abot a n accessible distro or OS? My previous replies were about Linux distros. If you talk about OSes in general, it is still another point: developing an universal OS? a layer able to apply to any OS (would be nearly research and dev, I know an european project trying this somesyears ago)?
>> There are several FOSS distros that target cell phones, but postmarketOS (pmOS) seems to be the only one which has the stated goal of keeping the old ones alive. Better yet, they are making serious progress at supporting a variety of old cell phones (mostly aarch64) on their (Alpine-based) Linux distribution. For details, see https://wiki.postmarketos.org/wiki/Devices.
>> OK, you say, but Alpine isn't my favorite Linux variant; I want to run X, Y, or Z. Well, that's why VMs were invented! Using something like QEMU, it should be possible to host almost any FOSS distro on top of pmOS. (Of course, if the guest distro is already ARM-based, that makes things easier and more efficient.)
> Do you know Cosmo Communicator? they do a high effort to have a phone with Android, Linux, etc. I think we can do something on such base.
>> In summary, I'm wondering what it would take to get versions of the more accessible OS distros set up to run (via a VM) on pmOS. Does anyone else think this is a feasible and/or worthwhile goal?
> I think debian is the most opened ecosystem to do this: this distro has an accessibility experience, persons to give opinion, we only lack of resources to port and maintain other architectures or implement improvements in the installer. Adding the hardware question is interesting but very complex, but... who knows
> Best regards