Waste of resources to ship your own OS?


#21

Screw FSF approval.
Purism is about making a machine that is owned by their owner. This is a much weaker condition than what the FSF wants. E.g. a software with a license that allows reading it but not passing on modified is anathema for the FSF, but for Purism’s goals, it would be fully sufficient.
Of course an FSF-“free” distro would be better, but rolling your own distro is going to be very expensive over time. Collaboration with an existing distro, or just acquiring the critical mass to attract enough people that it gets maintained by a community, those would be smarter moves IMHO.

BTW “pushing upstream” is really easy, at least for drivers. Just submit your patches to the Linux kernel people, distros will soon pick them up.
While waiting for the patch to make the roundtrip Purism -> kernel -> distro -> PureOS, the driver can be written as a loadable kernel module as an interim solution.


#22

Shh, don’t say things like that too loud or we’ll get the paddle…

But yeah I wasn’t aware that FSF approval essentially meant the device needs to be read-only or anything like that. Is that what you’re implying? Unless I’m reading your comment the wrong way here.

That makes sense for things like devices with free firmware, but I don’t think that’d make sense for a laptop… of course the drive can’t be read-only, that’d be ridiculous. I assume they must mean everything BUT the hard-drive being read-only, like the BIOS / Bootloader and stuff.

I wouldn’t WANT a computer that has a read-only hard drive. At least not as a personal computer. That sounds more like an… “access terminal” than a personal computer.

It actually sounds like a cool idea to me if I ever have the money to have two computers, one for just surfing the web and one that’s airgapped for everything else. But right now I need an all-around machine and something that’s read-only wouldn’t be good for that.


#23

@toolforger, you don’t own a device if you can’t modify it. This goes for the OS as well as a boot loader.
@Alex I think you missunderstood. FSF does not require such a thing.
Not even with the TPM module that is now available is anything read-only. It just means that only you, who has the master key, is allowed to modify it.

Being able to modify firmware: freedom
Others not being able to do so: security


#24

Nonono, I didn’t mean to say that the FSF requires read-only hardware. Quite the opposite actually.
The FSF demands that you can modify the software. I.e. that you get the sources, that you get the tools required to build them, and whatever else it takes to actually build it after modifying it. And that you can redistribute the new sources.
My point is: For the mission of Purism, it would be “just enough” if all software were “shared source”, i.e. if you can read it. In that situation, you could still verify that the software does nothing that restricts the user, and that’s the mission of Purism.
Now I’m evading the paddle by saying that FSF-approved would still be better; it’s just not (currently) practical.

Another point why one could ignore the FSF: It disapproves of all distros that merely allow the user to choose non-free software. However, that would be a non-issue for Purism: Just don’t install the non-free repositories. And, maybe, actively look for free alternatives. Though that’s hard to do in a world of patented Videocodecs - a distro that cannot play videos may be free, but it’s not useful for all tasks.
Here again a non-free codec would be against the FSF’s endorsement, but it would fit with Purism: Just code the thing up, cough up the licensing fees, publish the sources, people can validate the sources and experiment with them. They could even modify the code for experiments - they wouldn’t be allowed to publish the modified code, but they could publish the patches.


#25

If you cannot build it how can you be sure what’s running on your phone corresponds to what has been shared as source?


#26

Here are some interesting thoughts by “competition” (Pop!) why it makes perfect sense to create their own OS. These are just in addition to the reasons I stated earlier (also, I take back what I said on Pop! before :wink: )


#27

@ruff I didn’t say that you need not be able to build; in fact it is necessary for exactly the reason you state.
However, you do not need the ability to redistribute if you want to validate.

@Caliga sorry, no time to hear a full hour of video just to guess which of their points you want to make.


#28

You do realize that you wasted more than an hour of combined time of other users trying to convince them that Purism should be… less pure?
And then you can’t even realize that I elegantly crafted the link in a way that let’s you skip all the other intersting stuff they talk about, for your convenience?
:roll_eyes:

But as you do understand the concept of the value of time, you might also understand why Purism staff did not participate in this thread.
It’s pointless to try to convince people that the goals of Purism are (not) exactly right, just the way they are.


#29

@Caliga feel free to ignore me if you can’t stand running out of arguments - because that’s the usual reason why meta-arguments start.
Oh, and you’ve been misunderstanding my position anyway, but I’m not going to waste my time arguing that. If you want to put FSF above everything else, that’s fine by me, I can respect that - provided you can respect my position that the FSF’s arguments aren’t always the only thing to consider.


#30

Doing the FSF dance to get endorsed by the FSF might mean something, but I think the Purism brand has received a lot of exposure, so there is no need for Purism to go for FSF endorsement, although one could consider it a nice milestone.

After the milestone has been reached, one could simply have two distro images. One that is FSF endorsed and another which isn’t and meets @toolforger’s requirements (and 95% of users).

I can seen some value for a completely clean FSF approved system for highly critical systems, but at this point we are talking about a niche of a niche.


#31

My point is based on experience in dealing with some ‘exposed sources’ to qualify with GPL requirements but which are completely non-buildable. After receiving complain about license violation they publish something. But that something is completely useless, some relict linux tree snapshot with fragments of patches and no toolchain to build it whatsoever. So that kind of approach with “here, take your sources” is really meaningless.


#32

waste of resources to ship your own OS ?

by “your own OS” i assume you refer to PureOS.

PureOS is a GNU+Linux distribution that strives to obtain the FSF blessing.

Once obtained the FSF recognition is the highest standard of ethics in computing meaning Libre Hardware+Software.

How can any FSF philosophy loving GNU+Linux distribution be considered “your own” ? as if Purism owns PureOS.

FSF endorsed GNU+Linux Libre distributions concern the PUBLIC DOMAIN and cannot be called “your own” as you @cinderella have stated in the title.

PureOS beeing in the PUBLIC DOMAIN can only be called “OUR OWN” thus what is the nature of your question ?

what i consider a WASTE of RESOURCES is having close to 8 billion (ONLY) mobile devices that run proprietary hardware and software designed to ENSLAVE and OPRESS FREEDOM and manipulate REAL PROGRESS


#33

All of those points are easily addressed, with one word: Trisquel.

Trisquel is an FSF-endorsed distro that is:

  • 100% free;
  • actively developed;
  • privacy-conscious;
  • intended to be usable by non-technical people.

It is the distro offered by MiniFree on their FSF RYF-certified laptops (at the time of writing).

So, clearly, Purism does not need to spend resources on PureOS in order to achieve RYF certification.

Again, there does not seem to be a need for this. Replicant is a mobile operating system that is fully free and actively developed by privacy-conscious people.


Purism is commendable in working to ship as-free-as-possible hardware, with hardware privacy protections (e.g. killswitches that actually work) and with libre, privacy-protecting firmware, that supports existing fully free OSes. But wouldn’t it be great if it focused on that exclusively, instead of spreading itself thin on userland software? PureOS development necessarily involves duplicating efforts of existing OS developers (e.g. in skinning) and requires Purism staff time to be spent deciding unimportant issues. Making genuinely privacy-protecting hardware and firmware is already a lot for a small company to do, especially when customer service is taken into account. Purism’s progress - although impressive - could be faster and more robust without PureOS as a distraction.

Here is a great example of something that Purism staff would not be wastingspending their time or Purism users’ time considering, if they were just shipping Trisquel as the OS with a Librem-focused Heads/Coreboot version underneath.


#34

Yes, all of those points you picked. I had more in all the previous posts, for example here: upsteaming etc, GuiXSD.

Also, I would be surprised if the Trisquel, based on Ubuntu 16.04, would support the Librem hardware well.
And if Purism were to improve on that, they would effectively… fork it?

The point is, Purism is here to actively develop FOSS, and to advance it.
It is, of course, simpler to ship outdated hardware with a stable distro and call it a day.

Purism currently heavily contributes to the advancement of Wayland, Gnome and KDE (for everybody, not PureOS specifically).
And that is exactly what Linux hardware companies should do, no just exploiting existing stuff.

… will not work out of the box on the Librem 5. So, they will have to add drivers and stuff. And as upstreaming is slow, they’d end up copying (replicating) Replicant and add their stuff to it and name it PureOS.
So, that’s what they do. Just with a Debian base.

You seem to have a misconception here. They are duplicating Debian, not the effort to create it.
It’s like writing a foreword to an existing book, not like writing a book.
PureOS is 99,9% Debian testing main, plus some patches (most of which will be part of Debian at some point, due to upstreaming).
Explained quite well by Zlatan here: What is PureOS and how is it built?


#35

This whole thread could have been avoided had someone read up a little bit more on Purism. Their intentions are clearly laid out.


#36

This amounts to saying that all Linux hardware companies should also be (userland) software companies. I completely disagree with that.

Let Linux hardware companies do two things and do them well:

  1. tamper-evidently source, assemble and ship hardware that is compatible with Linux without requiring proprietary firmware, and that has hardware features to improve privacy (e.g. effective killswitches; sane chip choices; sane logical separation between chips at the PCB level);
  2. tamper-evidently source, assemble and ship privacy and security-orientated libre firmware to support their hardware (e.g. the way that Purism is working on Heads), and updates to this firmware.

This is sorely needed, and will, itself, help to advance the state of the upper layers in the stack. Why? Because it will enable more people to learn about and to obtain hardware that supports the privacy protections in those upper layers meaningfully. Contrast this with what happens currently: most people - due to lack of availability, lack of affordability, and lack of awareness about security, privacy, and libre hardware - run hardware that undermines those upper layer protections by requiring untrustworthy binaries, by lacking killswitches, and/or by giving modems direct access to RAM or suchlike.

I think you meant to say, “supporting” :wink:


#37

If purism wants to grow big then they need brand recognition. That is people think secure laptops and think purism just as you think high-end cars and think Rolls Royce. part of this is making their own OS, it allows them to set the standard that they need and to ensure that their products are accessible.


#38

Exploiting, as in “reaping what thy neighbor soweth”

as in “only use components that somebody else already added excellent drivers to the kernel and freed the firmware, dramatically limiting choices” ?

And then, how does this new firmware get to the customer? Hoping it finds its way into the one of the upcoming Trisquel releases? Delaying the new laptop line until it lands there?

Do you have any indication that a substantial part (other than settings, defaults) of the modifications to Debian in PureOS are in the upper layers?
My impression is that they focus exactly on the things you expect. Coreboot, deblobbling, firmware, drivers, energy-saving, standby, TPM, Heads, Nitrokey…
And of course it makes sense to integrate all these things, providing an out-of-the-box experience instead of maybe some tutorials how to set it all up.


#39

Wow… where to begin unpicking that?

Code, unlike crops, is indefinitely copy-able, so by using Trisquel, I don’t deprive Trisquel’s developers of the ability to use Trisquel.

I have to wonder what you think free software distro developers want people to do with their code. (Spoiler: typically, they want people to copy it and run it, in accordance with the licenses. They aren’t just building the distro for themselves.)

No. As I explained earlier, hardware vendors should ship hardware that is compatible, and ship libre firmware for it, too, as appropriate.

This does not necessitate relying upon core Linux (or *BSD) devs to have already added that support: the vendor or the manufacturer can do so.

If it is appropriate for inclusion in the Linux kernel, then it would reach Trisquel just as all Linux kernel updates do.

If not, then Purism, like any vendor or manufacturer, can distribute it themselves, just as they are doing here (although ideally, with bettery security than in that example).

Using PureOS, and reading threads on this forum, were the indicators that left me thinking that the PureOS efforts have gone into the sorts of things that new, derivative distros typically spend efforts on: backend infrastructure (repositories, etc), management (figuring out what they want to do and who to assign which tasks to), and unnecessary userland changes such as to look-and-feel.

Debian is already blob-free (fully-free “main” repo and Linux kernel since “Squeeze” around January 2011), so de-blobbing parts of Debian seems unlikely to be part of PureOS development.

(If, OTOH, you mean that the PureOS developers are deblobbing things from outside Debian for inclusion into Debian, that’s great, but it does not require PureOS to exist.)

Coreboot is not part of Debian, so Coreboot work seems unlikely to be part of PureOS development. Ditto Heads, and ditto, to some extent, firmware more generally, standby, and energy-saving. So, this work doesn’t require PureOS to exist, either.

That doesn’t leave much from your list. TPM? Nitrokey? Not sure that PureOS is doing, or needs to do, much in relation to these at the OS level. But here’s the point: whatever they might be doing in these respects, it is not something that requires the creation and maintenance of a whole new distro. It is the latter that I am arguing against, not the former.

This would be entirely possible by shipping the Librems with an OS pre-installed and minimally configured. It does not require a whole new distro.


#40
  1. “waste of resources to ship you own OS” - completely missed the point.
    PureOS is based on Debian Testing - it is as such a PUBLIC DOMAIN not a private venture in OS development.
  2. Debian is one of the few GNU/Linux Distributions that is more than 20 years old. stable and proven as a reliable OS. Debian,RedHat and Suse are THE 3 MAIN pillars of open code. Purism has done a great decission to derive from one of those 3 great branches unlinke many distributions that chose ubuntu which is a based on Debian(as such it is a tertiary branch more prone to break and be forgotten)
  3. Debian has two flavors one that IS RYF compatible in that it doesn’t ship any non-freedom respecting code and one that allows such code to run by DEFAULT. PureOS is based on the first.
  4. Purism has chose Debian as a development-back-bone for PureOS because it need to make SURE that this OS will have a ROCK solid and assured future and not be some overnight sensation that has yet to stand the test of time. we see many come-and-go Linux-distributions simply because they can’t stand the test of time.
  5. PureOS runs by default the default vanilla GNOME shell. Fedora,RedHat,CentOS also ship by default with GNOME so it is good to have a comon BASE.
    overall more is gained than what is beeing lost so this IS A GOOD CALL !