10 years of updates for chromebooks, including current devices

[Donut417]

11 minutes ago, E-waste said:

I think Google can do more, by explaining that a chromebook is just open-source Linux underneath, and also is compatible with the Google Chrome browser.  This would completely eliminate ALL software support timelines for any chromebook model.  I'm sure there is a way to lockdown Linux and prevent outside app installation, especially with a password enabled, that would probably be enough to keep the system functioning. 

The issue is the management is not done at the device level. Its done on the account. The first Google account logged in to the Chromebook determines whose the Admin/owner. However any other Google account can log in and have what ever privileges that account has.

 

Also not all those Chromebooks are the schools property. My Niece and Nephew both have their own. Their parents Google accounts is used as the Primary and they log in to their schools when they need to do school. I can tell you their dad nor my sister has a fucking clue when it comes to Linux. So expecting parents to know about that is asking a lot.

 

Part of your problem is you think everyone is tech savvy. They are not. Im a millennial, the generations before us have not a fucking clue when it comes to tech. The generations after ours can use tech but most of them don't understand how this stuff works. None of these young kids had to use a Windows Start up disk to prep the drive for Windows installation. None of these kids have had to use the command line to install stuff. Trust me, I dont care how user friendly of a distro you install, you WILL have to go to a terminal at some point.

[igormp]

1 hour ago, E-waste said:

It is my limited understanding based on tech podcasts like the Destination Linux / Hardware Addicts, that RISC-V will help solve this issue and make it easier for operating systems to support various cpu models, without specific compilations per model.  That would significantly improve the ease of support of Android and custom software like LineageOS or stock AOSP across various devices.

Either you understood it wrong or your sources were wrong.

RISC-V is just an ISA, pretty much like ARM. The main difference is that ARM can either sell you an existing µarch design of theirs (what most manufacturers go for with those Cortex X3s, M0s, A710s, etc), or sell you a license to do your own (like Apple did for their Mx lineup), while RISC-V is royalty free and you can build your own design without paying anything to anyone.

 

The issue with support for various CPU models is not related to the core µarch, most phones actually have the exact same cores in use, but rather the peripherals used, which have NOTHING to do with the ISA, so the problem would still be exactly the same if you were using a RISC-V device with custom peripherals that the manufacturer didn't give you sources for.

 

RISC-V actually brings another issue to the table: since you can do your very own design, you can add custom instructions and whatnot, which might require some special compiler changes and a special kernel to make use of it. Also, there's no requirement for the core you made to be open source, it can be as closed as you wish.

[Kisai]

3 minutes ago, igormp said:

RISC-V actually brings another issue to the table: since you can do your very own design, you can add custom instructions and whatnot, which might require some special compiler changes and a special kernel to make use of it. Also, there's no requirement for the core you made to be open source, it can be as closed as you wish.

Bingo, which is why I don't really see it being adopted for "general purpose OS", but more likely SBC (single board computer) solutions. eg robotics, motion capture, automotive computers, surveillance systems, etc.

 

We're actually getting farther away from "the year of linux desktop" because we keep trying to "make things easier" by hiding things in (HTML) menus. I very much doubt "hey, so if you want to use this SBC with this hardware platform, you're going to have to compile this custom kernel, with these these driver module firmware blobs...", no instead you're going to be buying closed solutions, that even though it might be running Linux, it will be like the WRT54G all over again, where if there is enough people interested in keeping that platform "open" then the open source people will try their hardest to keep it open so that that OS can run on different SBC's. But that doesn't mean every SBC can run a full featured OpenWRT distro, since not all the wireless routers have enough RAM or onboard storage.

 

https://openwrt.org/docs/techref/hardware/soc

 

Funny enough

https://www.cnx-software.com/2023/07/04/risc-v-router-dual-gbe-dual-usb-2-0-can-bus-rs485/

First known Risc-V SBC, the image itself shows the SoC plus two Realtek chips for the network.

 

*When I say SoC, I mean the CHIP, when I say SBC, I mean the SoC+PCB+connectors

 

I actually think in about 10 years, the idea of even installing an OS will be so foreign that it will feel like the COBOL-Y2K issue. People won't try to fix the software on completely useful device. On one hand, I like that I can let my device update the software. On the other hand, i hate that software developers can push an update that bricks the working version of the software, or shut down some cloud server, and thus I can no longer use it, and since I got blindsided by it, I can't just restore a backup of that software from before they pushed the poison pill version.

[igormp]

10 minutes ago, Kisai said:

Bingo, which is why I don't really see it being adopted for "general purpose OS", but more likely SBC (single board computer) solutions. eg robotics, motion capture, automotive computers, surveillance systems, etc.

Nothing stops it. Alibaba build actual server-grade CPUs to run linux on it.

If you meant as in general purpose for the wider consumer market, then I can agree with that, there's not much traction going on for a high-perf RISC-V µArch, which is understandable when the ISA lacks some more relevant vector stuff.

10 minutes ago, Kisai said:

unny enough

https://www.cnx-software.com/2023/07/04/risc-v-router-dual-gbe-dual-usb-2-0-can-bus-rs485/

First known Risc-V SBC, the image itself shows the SoC plus two Realtek chips for the network.

Wdym by first known? I've had a risc-v SBC in my own hands for over 2 years now lol

https://www.cnx-software.com/2021/05/20/nezha-risc-v-linux-sbc/

On 1/14/2022 at 12:34 PM, igormp said:

Is it really the first? I have a nezha D1 sitting right next to me.

 

Most linux distros should work, the risc-v support is reasonable.

 

[Kisai]

2 hours ago, igormp said:

Nothing stops it. Alibaba build actual server-grade CPUs to run linux on it.

If you meant as in general purpose for the wider consumer market, then I can agree with that, there's not much traction going on for a high-perf RISC-V µArch, which is understandable when the ISA lacks some more relevant vector stuff.

Yeah, like I don't see a "RISC-V Linux Desktop/Laptop" ever becoming a thing unless there is some wider agreement to not fork the ISA, and to only implement that which is agreed to by all vendors who wish to use "RISC-V" in marketing. eg, kinda like how "IBM Compatible" was essentially meaningless, but because every x86 CPU (eg NEC, Cyrix and AMD) supported the same core instructions, software that was written to only use the Intel instructions and never the extensions made by the clone chips. The big problem with open standards is the cat-herding, so progress is just slower than when a single vendor controls the standard and just licenses it.

 

You can build stuff that is pretty much closed off to the end user tampering with it though. It'll work, but it's not great from an e-waste perspective.

 

2 hours ago, igormp said:

Wdym by first known? I've had a risc-v SBC in my own hands for over 2 years now lol

https://www.cnx-software.com/2021/05/20/nezha-risc-v-linux-sbc/

 

The word "router" got dropped in that part of the comment, but the idea was that what we're mostly seeing are special purpose SBC's. The one I linked is basically something you'd use for a firewall/router. The one you linked appears to be something closer to an IoT platform.

 

At any rate, we seem to still see the same problem, proprietary IP blocks like the GPU. So you couldn't just yank the flash media from one RISC-V and stick it in another made by someone else.

[igormp]

33 minutes ago, Kisai said:

Yeah, like I don't see a "RISC-V Linux Desktop/Laptop" ever becoming a thing unless there is some wider agreement to not fork the ISA, and to only implement that which is agreed to by all vendors who wish to use "RISC-V" in marketing. eg, kinda like how "IBM Compatible" was essentially meaningless, but because every x86 CPU (eg NEC, Cyrix and AMD) supported the same core instructions, software that was written to only use the Intel instructions and never the extensions made by the clone chips. The big problem with open standards is the cat-herding, so progress is just slower than when a single vendor controls the standard and just licenses it.

I don't see this as much of an issue since you do have the base ISA, and as long as you properly implement it, the only thing you may be missing are extra proprietary extensions added on top, but the standard stuff should work fine. 

The problem I see is just lack of interest in such device. Not like a "Linux-only" laptop has any interest to begin with, no matter which ISA you use. 

35 minutes ago, Kisai said:

At any rate, we seem to still see the same problem, proprietary IP blocks like the GPU. So you couldn't just yank the flash media from one RISC-V and stick it in another made by someone else.

Even if you had a standard GPU, peripherals are pretty much the issue much like you see with ARM devices.