Does the crapware really pay more than the cost of a Windows EOM license?
How does that work economically? Because not everybody buys the crapware license later (and they normally cost less than Windows), and mass stealing of data is not very lucrative.
Typically OEM's are paid per install of crapware (regardless if the user ends up paying for the subscriptions).
It's just a form of marketing for these companies - ie. a certain percentage of potential users will convert into paying users.
The reason Windows laptops have been cheaper or the same price as the linux counterpart is usually due to the combination of subsidized licensing from Microsoft, crapware installation kickbacks, and lower cost of support infrastructure (the OEM already has a huge Windows support staff and infrastructure, creating a new one for Linux - or more correctly, a specific distro of linux - costs more).
If the linux laptops sell more and more, eventually the support costs will come down to on-par (or lower perhaps... who knows). It's a chicken-and-egg problem, and one that OEM's are not willing to foot the bill up front for. We have to vote with our wallets and take that short-term hit in order to push the ecosystem in that direction.
But crapware can not pay that much. The crapware creators themselves get less revenue than licensing costs, and they can kick back just a fraction of that revenue. (That is, unless they have another business model that isn't selling licenses or spying on users.)
Even say, $1.00 per system sold with crapware pre-installed can make a difference in the profitability of a product line at scale. Then there's the kickbacks for when a customer does pay for a subscription.
When I worked at Corel (many, many years ago), I was stunned by how much we paid to put Word Perfect on OEM computers. It certainly wasn't worth it from the perspective of a single sale. I think we easily spent more than we recouped from sales. The thinking was that the person who bought would also upgrade, though, and I think that was sound reasoning. Vendor lock-in is a powerful force. Personally, I thought it was a waste, though. Cost of sales was a large multiple of development cost... :-P
The correct comparison would be "Linux + Windows" one will have less margin than "Windows only" one, only because of cost of testing / opportunities of bundling software with it.
The same could be said about windows on supercomputers. Why is windows supported so poorly by Cray and other HPC supercomputer vendors when Linux is supported so well?
Actually, MSFT has pushed Windows Server on supercomputers a few times, as well as for other latency critical applications like the London Stock Exchange. Each time they've ended up with egg on their faces for technical reasons, i.e. Linux consistently outperforms Windows Server both on speed and reliability for these workloads.
Right, you're simply providing reasons that agree with me!
I work as a tech guy (linux wrangler / code monkey) for a finance firm. I worked in finance during the meltdown of LSE and the new linux based "Millennium" platform. I agree with you that Linux consistently is higher throughput AND lower latency, but that is my point. Windows isn't on supercomputers because it isn't a great fit and there isn't much of a market for it. Ditto can be said for Linux on the Desktop.
* This message brought to you from my RHEL7 work workstation running GNOME 3 and Chrome :D
This may be slightly off-topic (and it's admittedly an older incident), but the fiasco with the Aegis missile cruiser, the USS Yorktown, often comes to mind when I read of high-value or mission-critical systems using Windows of any kind. Sadly, the Yorktown is certainly not unique with respect to major projects relegated to using Windows OSes- it just happens to be a particularly vibrant example of the potential results.
Huh, 21.7% is a surpizingly huge portion in my book. I would have suspected a much smaller number.
One fifth is a market segment that is large enough to cater for. If you even put a small team on the task of ensuring that Linux installs on your notebooks, and make the minimal neccessary changes (like this BIOS change... or better revert of a breaking change), then you could claim Linux support as your unique selling point, and grab a huge fraction of those 20%.
>One fifth is a market segment that is large enough to cater for
Keep in mind that that's 21.7% of stack overflow users (i.e. developers), not users in general.
According StatCounter (based on browser stats), GNU/Linux represented 1.44% of desktop/laptop OS as of last month. Perhaps developers are more likely to use these laptops in particular, but they're unlikely to be 100% of users.
> Keep in mind that that's 21.7% of stack overflow users (i.e. developers), not users in general.
The unfortunate reality is that while this number is largerish, whenever a company comes out with a linux-based "developer edition" laptop, folks don't buy it.
The Dell XPS 13, for example, is a great looking machine - but too many folks were busy buying macbooks to help Dell learn this market segment does indeed want more products like this.
We need to vote with our wallets, in order to reinforce this behavior.
Perhaps this may change now that developers are feeling disenfranchised with Apple over the recent model releases.
> whenever a company comes out with a linux-based "developer edition" laptop, folks don't buy it
> The Dell XPS 13, for example, is a great looking machine - but too many folks were busy buying macbooks to help Dell learn this market segment does indeed want more products like this
Not sure about the why other (potential) Linux users might not buy notebooks such as this, but despite single-booting Linux for the past several years, I wouldn't be any more likely to buy a computer with Linux pre-installed on it because the chances that the computer has a distro/configuration I would want to use is basically zero; if I'm going to have to reinstall Linux and/or configure it from scratch, then the fact that it's specifically marketed as a Linux laptop doesn't really give me anything. I suspect that this might be at least partially responsible for "developer edition" laptops failing, as at least in my anecdotal experience, many people who are looking for an out-of-the-box experience aren't generally inclined to use Linux as their main OS in the first place.
> the chances that the computer has a distro/configuration I would want to use is basically zero
Couldn't agree more (as a Fedora user myself! - Ubuntu is typically the "blessed" distro on these systems for various reasons).
However, "linux support" is more than a chosen distro - it's the supporting code necessary to make linux a first-class citizen on the given hardware (driver support, setup scripts, porting popular software, etc...), as well as support knowledge and training at the OEM (bash is bash, and Firefox is Firefox on Ubuntu, Fedora, Arch or whatever).
I mentioned this on the sibling comment, but I've honestly never had any issues with drivers/hardware support with Linux on laptops ( with the exception of putting Linux on an early 2015 Macbook shortly after they came out, which had a new wifi card model, one of those new-at-the-time force touch trackpads that used different drivers, etc.). That being said, I only started using Linux in 2012, and I know that things used to be much worse in that regard.
As for technical support and training, I can see that being useful in an enterprise context, but again, I honestly can't see the type of people who want to single-boot Linux on their laptop being interested in support and training from Dell, HP, etc. It's been my experience that by the time someone is comfortable enough to use Linux as their sole OS (which many people don't want to do despite being comfortable enough due to needing Windows for other reasons, like gaming), they aren't really going to need or want any tech support from an OEM.
Dells are great as far as Linux support. I have an Inspiron at work and a Dell XPS 13 at home, but run Linux natively. I was kinda surprised everything worked on the first try .. Wi-Fi, Bluetooth, extra displays on the docking station .. the works.
I was more talking about software (OS-specific and DE configuration, packages, etc.) rather than hardware. I've only first started using Linux about four years ago, and I understand it used to be quite different, but I've honestly never had any significant trouble with regards to hardware working on Linux. With the exception of the time that I put Arch on an extremely new Macbook, everything has always worked pretty well out of the box for me. The only hardware-related thing I've ever had to configure on my non-Apple laptops is adding a kernel parameter in the bootloader config to get the backlight adjustment to work.
On the other hand, I've yet to find a distro that pre-installs a GUI with configuration and packages anywhere close to what I want in a setup. I go out of my way to use distros with an option to install without a desktop environment even if it has an option to install the one I want, since it usually comes with a lot of extra crap I don't want. Obviously I'm probably a little more extreme here than even most Linux users, but at least among Linux power users (i.e. the people I'd expect to be interested in buying a laptop to single-boot Linux) that I know, Unity is considered to be kind of terrible.
Am I wrong in assuming that these types of "developer edition" laptops will generally come with a vanilla Ubuntu install? I admit that I've never really looked into getting one, so it's quite possible I'm mistaken in understanding the kind of setup they provide
I bought an XPS15 and I put Linux on it. But I didn't buy the "developer's edition" because I still wanted a Windows install, that pretty much just sees gaming. So I don't show in the statistics for Dell as an obvious Linux user, but if my research had turned up that the XPS15 worked poorly in Linux, I would never have bought it.
Almost all Linux users are familiar with the process of putting Linux on a machine, since that's the only way to get it. The thing that's primarily helpful about seeing a manufacturer have a "blessed" linux system to me is more that I can be confident all the drivers I need exist; I'd probably whack the install they sent me and replace it with something else anyhow.
The problem is, generally the only thing that "Powered by Linux" laptops offers (I am talking about big manufacturers like Dell here, not System76 or others) is Linux pre-installed.
Those notebooks are generally more expensive than their Windows counter parts (because Microsoft subsides then) and does not offer any customization (Dell XPS 13 Developer Edition still comes with a "Windows" key, wtf?!), so why bother?
In my case, coming with either Windows or Ubuntu is exactly the same, since I use Arch Linux and will have to reinstall anyway.
To back up a step, why should there even be a "linux developer edition" model? Why can't they just make a solid laptop to begin with, that also just happens to support Linux? What's honestly different? The icon on the meta key?
To answer your question, as a picky hardware nut I have yet to see anything worth paying extra for. Give me a 4lb 15" 1080p matte IPS with user-expandable RAM and a PCIe m.2 slot, a SATA drive bay instead of a cdrom, _no_ dedicated graphics card, a non-chiclet keyboard, a high quality trackpad _centered under the space bar_, and a user-replaceable (hot-swappable?) battery that gets 8-10hrs on a single charge and I'll think about forking out extra money. Otherwise eff off, I'm happy with used Thinkpads.
It's a tough crowd to please, to be sure, especially when half of us are perfectly capable and willing to put in the extra effort to get random finicky component X working.
Yes, I'm well aware. My point was that I and many others would happily pay extra to pad their margins if anybody would just make a laptop screen/chassis/motherboard barebones combo worth paying for. In my judgment, nobody has understood the audience well enough to convince the business types to let them even try to actually get a truly professional-grade machine to market.
And when I say professional-grade, I mean sturdy, expandable, and hackable to the core. Designed for Linux/*BSD first, that still runs Windows if the end user wants to install it themselves. If something like the Novena can succeed with a 1.2GHz ARM, I'll be damned if something similar with a quadcore x86 wouldn't do a hell of a lot better. It may bring in small change to a Dell of the world, but there has to be _somebody_ in China who would think otherwise.
The way to do it, I think, would be through a crowd-funding type of platform. 3D print a prototype, prove the sales are there, and then you can spend the money on design and tooling for manufacturing.
Lately my impression is that Mac is still gaining ground with developers. At least where I work, virtually every new developer asks for a Mac. I find it a bit strange because then they have to faff around with homebrew to get a decent GNU build, which is what we need for development. Pretty (Mac) still seems to be favoured over flexible (GNU/Linux). Maybe one day...
My gut feeling is that 0.5% of PCs/Macs sold for business use are running desktop linux in the UK. This is based entirely on my experience of working for internet based businesses over the last decade. Where I am right now that figure is actually true, we have 200 or so PCs/Macs and my box is the only linux one. Okay, anecdotal but ballpark.
Lenovo have a 20% market share in the UK and my gut feeling is that 20 million UK employees sit in front of a computer rather than a sat-nav or a point of sale system etc.
Given that Lenovo have dozens of Thinkpads and Yoga variants, there are probably a dozen machines of interest to the linux desktop type of person.
If we assume that PCs/Macs get replaced every two years (generous) then the arithmetic works out at around a thousand sales per model per year to desktop linux users. I run normal Ubuntu not some variant - Mint/Debian/etc. Linux != Ubuntu, there are those that prefer the Red Hat flavours. Therefore a default Linux install of Ubuntu isn't going to be what picky linux devs will want.
Note that in the UK the keyboard is different and there is a 'u' in colour. You would want your default linux install to work out the box without having to change locale settings.
Obviously the US/Canada market is vast compared to the UK, but not more than 10 times larger. So that still means a low figure for expected linux sales even for the gigantic Lenovo company. Why would they bother to support it? There is no business case.
> This is a Linux-only BIOS, meaning it should be used only by those who want to load the OS. If you want to continue with Windows, do not load the firmware. “This BIOS is not intended to be used on machines running Windows operating systems,” Lenovo said.
The "Linux only BIOS" also works for Windows though so you can dual boot with it. The reason why they say it doesn't is because they don't want to spend money on officially supporting it and AHCI gives a slight performance decrease from their default RAID configuration.
Well if it does work with Windows, then fine, I took it from the article, that there's no chance running Windows or dual booting after switching to it.
> because they don't want to spend money on officially supporting it
I hope this doesn't mean the warranty and support is void after doing this, wouldn't even surprise me.
> I took it from the article, that there's no chance running Windows or dual booting after switching to it
That's literally what they said. The GP is telling you it's a lie (and is very probably right, but I don't know the details), but they said you can not dual boot.
> "and AHCI gives a slight performance decrease from their default RAID configuration."
Is that actually true, or is it really that using Microsoft's AHCI driver instead of Intel's driver hurts performance? Last time I checked, Intel's driver could be used even when the controller is in AHCI mode.
A lot of the newer developer grade laptops have nvme options anyway (which work on both Windows and current Linux kernels). My XPS 13 runs on Linux with nvme fine.
I suspect you're correct and that the new BIOS may break Windows certification. My experience is that Windows 10 runs fine on non-certified hardware since it's running fine on four non-certified machines here at the house.
> Locking specific hardware to Windows has emerged as an issue in recent years. For example, Intel’s Kaby Lake chips support only Windows 10, so will AMD’s upcoming Zen chips.
> Another option is to buy Linux-based laptops with the latest Kaby Lake chips, like System76’s Lemur laptop, which starts at $699, or Dell’s XPS 13 Developer Edition, which starts at $949.
So, which is it? Can Kaby Lake chips support Linux or not, because those statements seem contradictory.
Misleading line. They do not support any version of window before 10 (that is, neither MS nor Intel or AMD will release drivers for other windows). Linux (and other OSs) should support them just fine.
I'm curious, what do the CPUs need drivers for? I thought x86 processors are always backwards-compatible, in theory all the way back to DOS.
Or is it the chipset driver? Here also I wonder what would have changed dramatically since the previous generation.
I've found the only thing I needed to install chipset drivers for was for mainboard-specific features, like LED and Fan control, and for (back then) bleeding edge features like USB3. Granted, this was because basic chipset support was integrated into windows. Still, I wonder what the breaking change or new feature in Kaby Lake is.
That's what I gathered. It just means older operating systems won't be able to take advantage of new instruction sets, or as other commenters have pointed out, we won't see USB3/Bluetooth drivers that are guaranteed to work on the older chips.
It's the "chipset" drivers, yes; but in the modern world all that hardware is on-chip and not meaningfully separable from the SoC. So sure, x86 code will run fine on the chip in any OS (e.g. Linux in virtualization under Windows 10). But you still can't boot Linux on it without drivers for the hardware.
In this case it's the SATA storage controller, which is configured by the BIOS in a mode that isn't compliant with the AHCI spec.
> In this case it's the SATA storage controller, which is configured by the BIOS in a mode that isn't compliant with the AHCI spec.
Not quite. It's the PCIe bridge, configured in a thoroughly nonsensical way that makes it pretend that an NVMe device (which is a PCIe device in its own right) is hidden behind a SATA controller.
Most ordinary Intel Core CPU's incorporate a graphics engine with multiple cores. So under Linux, they can use proprietary graphics drivers. With a discreet GPU, the story becomes even muddier, because Intel's Optimus (a technology that switches between discreet and on CPU graphics for power savings) is not fully implemented by Nvidia's Linux drivers. So there's no dynamic switching like there is in Windows.
Yes, I noticed that too. I wonder if the authors of these articles care about what they are writing.
Kaby Lake works with any OS which has drivers for them. MS didn't plan to make any for Windows < 10 and wasn't asking OEMs to develop them for those Windowses.
I see, so the first statement was made in a context of only the windows ecosystem. It's exceptionally poor wording, since they aren't actually talking about "Locking specific hardware to Windows" but instead "ensuring specific hardware doesn't get support on older version of windows". Windows 10 and windows 8 and prior are all still "Windows".
Too late, I will never buy Lenovo again, there are too many other reasonable choices of companies who don't make anti-Linux decisions then backtrack when they get covered in shit. Once bitten..
I have a lenovo thinkpad x140e and I love it. It's underpowered and "bulky", but as a healthy adult I really don't care if a laptop is a few hundred grams heavier. I've dropped it once or twice as well.
But I am not going to buy a laptop that won't run Linux. have any recommendations?
I have been thinking about whether the distinction between "consumer" (eg IdeaPad) and "business" (eg ThinkPad) machines even makes sense. This is not even limited to Lenovo of course.
Nope, because Intel can just push some binary drivers out and everyone will down it with a gargle.
The kernel patches are refused because the kernel devs expect to maintain the code for as long as there are devices in active use that need it once allowed in.
Thus they demand a proper spec for the device behavior so they can test for breakages down the line. And apparently the patches are already violating AHCI behavior expected by the kernel.
Or at least that is what i got out of reading the short ML thread.
I'm not too familiar with this area of PC hardware, but it does seem like their AHCI controller can be configured to act like a PCI bridge with an NVMe device on the other side of it.
I read it and even tried to search for his name because apparently he has made important contributions to Linux (he links /usr/src/linux/CREDITS as his CV so...), but couldn't find anything specifically about his work.
Do you happen to know what does he work/has worked on regarding Linux?
This is interesting, considering I'm running stock ubuntu 16.04 on a lenovo yoga 900. Didn't have to do any magic at install. I was under the impression that newer kernels already supported The Thing That Is The Problem. It's either that or my laptop is magic...
At least it's something, but good luck getting any refund from them for the Windows tax. Unless you summon them to a small claims court, they staunchly refuse to refund you, even if you don't plan to use Windows and don't accept the Windows EULA.
Wait, why would they? The Windows license is bundled with the cost of the computer. You bought the license, whether you plan to use it or not. They are under no obligation to sell the license separately or offer you partial refunds, that's not how sales work.
Good on them if they do, but I feel like you'd be laughed out of Small Claims court. What's your argument? You weren't forced to buy a Lenovo computer, you could have bought any machine, including one with Linux preinstalled and no "Windows Tax."
Hrm! IANAL, I was entirely unaware this was even a thing, and now I'm deep down the Wikipedia rabbit hole. The more you know!
Looks like the legality of it varies by state / region / country. Maybe it's not illegal where I live; I'm in South Texas. I'll research it though, thanks!
https://en.wikipedia.org/wiki/Tying_(commerce)
So if I'm understanding this correctly, Lenovo isn't actively blocking Linux, they're just using a newer RAID mode in their storage controller that there are no existing Linux drivers for?
This doesn't seem like that big of a deal actually. It's nice of them to provide a legacy mode bios, but this will perhaps encourage the community to write open source drivers for the storage controller in its native mode, expanding the hardware that the linux kernel can support. That would be a win win.
Well, they initially only released a custom BIOS with a goto statement added to jmp out of the disk controller mode setting (which is normally present and allows the user to change back to AHCI mode). Reverse engineers on the Lenovo forums discovered this modification while studying the disassembly.
Then, one user patched the BIOS and manually reflashed it using an SPI flasher and some soldering, and Linux worked just fine by detecting the drive.
Yes, the ideal solution is that Intel gives specs to OSS devs to build a driver (or even builds on themselves), but the path of least resistance is to have Lenovo unblock that BIOS setting, which amounts to changing one line of code, or 15 minutes of one engineer's time.
101 comments
[ 0.26 ms ] story [ 2362 ms ] threadHow does that work economically? Because not everybody buys the crapware license later (and they normally cost less than Windows), and mass stealing of data is not very lucrative.
It's just a form of marketing for these companies - ie. a certain percentage of potential users will convert into paying users.
The reason Windows laptops have been cheaper or the same price as the linux counterpart is usually due to the combination of subsidized licensing from Microsoft, crapware installation kickbacks, and lower cost of support infrastructure (the OEM already has a huge Windows support staff and infrastructure, creating a new one for Linux - or more correctly, a specific distro of linux - costs more).
If the linux laptops sell more and more, eventually the support costs will come down to on-par (or lower perhaps... who knows). It's a chicken-and-egg problem, and one that OEM's are not willing to foot the bill up front for. We have to vote with our wallets and take that short-term hit in order to push the ecosystem in that direction.
But crapware can not pay that much. The crapware creators themselves get less revenue than licensing costs, and they can kick back just a fraction of that revenue. (That is, unless they have another business model that isn't selling licenses or spying on users.)
Even say, $1.00 per system sold with crapware pre-installed can make a difference in the profitability of a product line at scale. Then there's the kickbacks for when a customer does pay for a subscription.
Unless, of course MS is paying the OEM to install Windows, or billing for Linux computers. Or crapware has some business model that I'm not aware of.
It is just about the demographics.
Example: http://www.computerworld.com/article/2467082/data-center/lon...
I work as a tech guy (linux wrangler / code monkey) for a finance firm. I worked in finance during the meltdown of LSE and the new linux based "Millennium" platform. I agree with you that Linux consistently is higher throughput AND lower latency, but that is my point. Windows isn't on supercomputers because it isn't a great fit and there isn't much of a market for it. Ditto can be said for Linux on the Desktop.
* This message brought to you from my RHEL7 work workstation running GNOME 3 and Chrome :D
https://gcn.com/Articles/1998/07/13/Software-glitches-leave-...
Mac OS X 26.2%
Windows 7 22.5%
Linux 21.7%
Windows 10 20.8%
Windows 8 8.4%
Windows XP 0.4%
Though it is possible that the sort of developers that use desktop linux don't use Stack Overflow.
[0] http://stackoverflow.com/research/developer-survey-2016#tech...
One fifth is a market segment that is large enough to cater for. If you even put a small team on the task of ensuring that Linux installs on your notebooks, and make the minimal neccessary changes (like this BIOS change... or better revert of a breaking change), then you could claim Linux support as your unique selling point, and grab a huge fraction of those 20%.
Keep in mind that that's 21.7% of stack overflow users (i.e. developers), not users in general.
According StatCounter (based on browser stats), GNU/Linux represented 1.44% of desktop/laptop OS as of last month. Perhaps developers are more likely to use these laptops in particular, but they're unlikely to be 100% of users.
The unfortunate reality is that while this number is largerish, whenever a company comes out with a linux-based "developer edition" laptop, folks don't buy it.
The Dell XPS 13, for example, is a great looking machine - but too many folks were busy buying macbooks to help Dell learn this market segment does indeed want more products like this.
We need to vote with our wallets, in order to reinforce this behavior.
Perhaps this may change now that developers are feeling disenfranchised with Apple over the recent model releases.
> The Dell XPS 13, for example, is a great looking machine - but too many folks were busy buying macbooks to help Dell learn this market segment does indeed want more products like this
Not sure about the why other (potential) Linux users might not buy notebooks such as this, but despite single-booting Linux for the past several years, I wouldn't be any more likely to buy a computer with Linux pre-installed on it because the chances that the computer has a distro/configuration I would want to use is basically zero; if I'm going to have to reinstall Linux and/or configure it from scratch, then the fact that it's specifically marketed as a Linux laptop doesn't really give me anything. I suspect that this might be at least partially responsible for "developer edition" laptops failing, as at least in my anecdotal experience, many people who are looking for an out-of-the-box experience aren't generally inclined to use Linux as their main OS in the first place.
Couldn't agree more (as a Fedora user myself! - Ubuntu is typically the "blessed" distro on these systems for various reasons).
However, "linux support" is more than a chosen distro - it's the supporting code necessary to make linux a first-class citizen on the given hardware (driver support, setup scripts, porting popular software, etc...), as well as support knowledge and training at the OEM (bash is bash, and Firefox is Firefox on Ubuntu, Fedora, Arch or whatever).
As for technical support and training, I can see that being useful in an enterprise context, but again, I honestly can't see the type of people who want to single-boot Linux on their laptop being interested in support and training from Dell, HP, etc. It's been my experience that by the time someone is comfortable enough to use Linux as their sole OS (which many people don't want to do despite being comfortable enough due to needing Windows for other reasons, like gaming), they aren't really going to need or want any tech support from an OEM.
On the other hand, I've yet to find a distro that pre-installs a GUI with configuration and packages anywhere close to what I want in a setup. I go out of my way to use distros with an option to install without a desktop environment even if it has an option to install the one I want, since it usually comes with a lot of extra crap I don't want. Obviously I'm probably a little more extreme here than even most Linux users, but at least among Linux power users (i.e. the people I'd expect to be interested in buying a laptop to single-boot Linux) that I know, Unity is considered to be kind of terrible.
Am I wrong in assuming that these types of "developer edition" laptops will generally come with a vanilla Ubuntu install? I admit that I've never really looked into getting one, so it's quite possible I'm mistaken in understanding the kind of setup they provide
Almost all Linux users are familiar with the process of putting Linux on a machine, since that's the only way to get it. The thing that's primarily helpful about seeing a manufacturer have a "blessed" linux system to me is more that I can be confident all the drivers I need exist; I'd probably whack the install they sent me and replace it with something else anyhow.
Those notebooks are generally more expensive than their Windows counter parts (because Microsoft subsides then) and does not offer any customization (Dell XPS 13 Developer Edition still comes with a "Windows" key, wtf?!), so why bother?
In my case, coming with either Windows or Ubuntu is exactly the same, since I use Arch Linux and will have to reinstall anyway.
To answer your question, as a picky hardware nut I have yet to see anything worth paying extra for. Give me a 4lb 15" 1080p matte IPS with user-expandable RAM and a PCIe m.2 slot, a SATA drive bay instead of a cdrom, _no_ dedicated graphics card, a non-chiclet keyboard, a high quality trackpad _centered under the space bar_, and a user-replaceable (hot-swappable?) battery that gets 8-10hrs on a single charge and I'll think about forking out extra money. Otherwise eff off, I'm happy with used Thinkpads.
It's a tough crowd to please, to be sure, especially when half of us are perfectly capable and willing to put in the extra effort to get random finicky component X working.
Hardware margins are thin, unless you're Apple.
And when I say professional-grade, I mean sturdy, expandable, and hackable to the core. Designed for Linux/*BSD first, that still runs Windows if the end user wants to install it themselves. If something like the Novena can succeed with a 1.2GHz ARM, I'll be damned if something similar with a quadcore x86 wouldn't do a hell of a lot better. It may bring in small change to a Dell of the world, but there has to be _somebody_ in China who would think otherwise.
The way to do it, I think, would be through a crowd-funding type of platform. 3D print a prototype, prove the sales are there, and then you can spend the money on design and tooling for manufacturing.
Every attempt I've seen, though, has either failed, or fallen prey to the lock-in-based lure of Microsoft.
There are other laptop lines targeted for developers.
I need to update my LinuxCounter [1] profile after years of neglect.
[1]: https://www.linuxcounter.net/
(and I should update my profile too)
The fact that there are about as many developers who are Linux users as there are Mac users is quite telling.
Lenovo have a 20% market share in the UK and my gut feeling is that 20 million UK employees sit in front of a computer rather than a sat-nav or a point of sale system etc.
Given that Lenovo have dozens of Thinkpads and Yoga variants, there are probably a dozen machines of interest to the linux desktop type of person.
If we assume that PCs/Macs get replaced every two years (generous) then the arithmetic works out at around a thousand sales per model per year to desktop linux users. I run normal Ubuntu not some variant - Mint/Debian/etc. Linux != Ubuntu, there are those that prefer the Red Hat flavours. Therefore a default Linux install of Ubuntu isn't going to be what picky linux devs will want.
Note that in the UK the keyboard is different and there is a 'u' in colour. You would want your default linux install to work out the box without having to change locale settings.
Obviously the US/Canada market is vast compared to the UK, but not more than 10 times larger. So that still means a low figure for expected linux sales even for the gigantic Lenovo company. Why would they bother to support it? There is no business case.
One step forward, two steps back.
> because they don't want to spend money on officially supporting it
I hope this doesn't mean the warranty and support is void after doing this, wouldn't even surprise me.
That's literally what they said. The GP is telling you it's a lie (and is very probably right, but I don't know the details), but they said you can not dual boot.
Is that actually true, or is it really that using Microsoft's AHCI driver instead of Intel's driver hurts performance? Last time I checked, Intel's driver could be used even when the controller is in AHCI mode.
https://mjg59.dreamwidth.org/44694.html
Apparently, there isn't much difference between AHCI and RAID modes in the hardware itself.
> Another option is to buy Linux-based laptops with the latest Kaby Lake chips, like System76’s Lemur laptop, which starts at $699, or Dell’s XPS 13 Developer Edition, which starts at $949.
So, which is it? Can Kaby Lake chips support Linux or not, because those statements seem contradictory.
Or is it the chipset driver? Here also I wonder what would have changed dramatically since the previous generation.
I've found the only thing I needed to install chipset drivers for was for mainboard-specific features, like LED and Fan control, and for (back then) bleeding edge features like USB3. Granted, this was because basic chipset support was integrated into windows. Still, I wonder what the breaking change or new feature in Kaby Lake is.
I suspect this is mostly artificial and the EHCI emulation is still there but just fused off to encourage people to upgrade.
Is that it, or is there more to this I'm missing?
In this case it's the SATA storage controller, which is configured by the BIOS in a mode that isn't compliant with the AHCI spec.
Not quite. It's the PCIe bridge, configured in a thoroughly nonsensical way that makes it pretend that an NVMe device (which is a PCIe device in its own right) is hidden behind a SATA controller.
Wait, wasn't Optimus Nvidia's baby?
Kaby Lake works with any OS which has drivers for them. MS didn't plan to make any for Windows < 10 and wasn't asking OEMs to develop them for those Windowses.
I remember that stuff like this used to quickly be redirected to /dev/null like five years ago.
Great news then! Something has changed then :)
https://www.engadget.com/2015/08/12/lenovo-updater-security-...
https://www.cnet.com/news/superfish-torments-lenovo-owners-w...
But I am not going to buy a laptop that won't run Linux. have any recommendations?
Just keep buying Thinkpads. They are fine and support Linux. I would stay away from regular consumer laptops anyway.
However, the kernel maintainers turned down the patch because the hardware itself was bizarre, underspecified, and unreliable in common cases.
This is an ongoing story and i expect there'll be a resolution in a few months.
ML discussion thread is here: http://lists.infradead.org/pipermail/linux-nvme/2016-October...
The kernel patches are refused because the kernel devs expect to maintain the code for as long as there are devices in active use that need it once allowed in.
Thus they demand a proper spec for the device behavior so they can test for breakages down the line. And apparently the patches are already violating AHCI behavior expected by the kernel.
Or at least that is what i got out of reading the short ML thread.
It sounds like trying to tunnel NVMe through something that is only connected over an AHCI interface.
Do you happen to know what does he work/has worked on regarding Linux?
More about him, here: http://www.advogato.org/person/dwmw2/
Good on them if they do, but I feel like you'd be laughed out of Small Claims court. What's your argument? You weren't forced to buy a Lenovo computer, you could have bought any machine, including one with Linux preinstalled and no "Windows Tax."
Because they are obligated to. Forced product bundling is illegal - it violates anti-trust law.
Looks like the legality of it varies by state / region / country. Maybe it's not illegal where I live; I'm in South Texas. I'll research it though, thanks! https://en.wikipedia.org/wiki/Tying_(commerce)
This doesn't seem like that big of a deal actually. It's nice of them to provide a legacy mode bios, but this will perhaps encourage the community to write open source drivers for the storage controller in its native mode, expanding the hardware that the linux kernel can support. That would be a win win.
Then, one user patched the BIOS and manually reflashed it using an SPI flasher and some soldering, and Linux worked just fine by detecting the drive.
Yes, the ideal solution is that Intel gives specs to OSS devs to build a driver (or even builds on themselves), but the path of least resistance is to have Lenovo unblock that BIOS setting, which amounts to changing one line of code, or 15 minutes of one engineer's time.