Windows Subsystem for Linux 2 To Offer Faster I/O Performance, Native Docker Containers

[sheldonl]

and by the time they arrive at WSL 3, Windows will end up being only a thin GUI layer running a top some Microsoft Linux distribution :-P

Since when is Windows a 'thin' anything?

[bpetty]

And with this news we finally enter a time many never thought would come, the Year of the Linux Desktop... its just that the desktop is technically Windows.
"This isn't the Desktop we need, its the Desktop we deserve."

[oiaohm]

And with this news we finally enter a time many never thought would come, the Year of the Linux Desktop... its just that the desktop is technically Windows.
"This isn't the Desktop we need, its the Desktop we deserve."

I would not say this is "the year of the Linux desktop" just yet. It "the year of Linux kernel on the Desktop" this is slightly different.

https://en.wikipedia.org/wiki/Cooperative_Linux

Really this project that died out with 64 bit signed drivers was well head in supported features to WSL v1 and could be still ahead of WSL v2.

I would class WSL 2 as start of "Year of the Linux Desktop" if it turns out graphical applications built as Linux applications can work perfectly. At that point those making desktop applications would be able to choose Linux and forgot making Windows binaries this would make quite a few open source projects very happy reducing their build server workloads.

Removing dominate market player can be a slow bit by bit process.

I am expecting WSL2 to solve most of the IO trouble and this to be replaced with my transcoding is slow and other items like it that require GPU access to perform well.

Yes even items like the postgresql database with different forms of acceleration need gpu access.

Microsoft is on a very slippery slope restricting to terminal does not in fact stop Linux applications at different times wanting gpu access. Once you give GPU access you fairly much will have graphical Linux native applications being option.

[nils_]

Could be a Hyper-V solution with a Linux kernel running on it. Would certainly have faster I/O. WSL is really slow as it is today.

That's precisely what it is, probably with some kernel glue to better integrate with the host (like the virtualbox shared folders for example).

[Guy1524]

WSL 2 is a much less interesting project in my opinion. It's essentially just running Linux in a VM in a way that is seamless with the OS experience. This has been possible for years and years. According to a slide I saw in a Microsoft presentation on the subject, the userspace applications are also within this VM, so I'm unsure how they will be able to interact with a windows X server.

I really hope they keep WSL 1 around, the tech behind it is just so much interesting and promising in my opinion.

Also, in the long term, WSL excites me because it will remove the need for Linux devs to need to port their applications to windows. They can just make a simple MSI installer for their application that makes sure the user has a linux distro and windows X11 server installed and then just ship their linux binary. This will be very convenient for linux developers, and decrease the total amount of win32 software circulating, which is essential if we eventually want to phase it out.

[mattlach]

Embrace, Extend, Extinguish

[techzilla]

I support what Microsoft is doing these days, in regards to opensource. They have done a lot of very solid improvements, in both their corporate culture, and in their outreach efforts. In addition Linux will not ever be able to create a standardized arm platform, if it were to lead, as we've seen the effects of Google's leadership on the arm platform.

Everything I loved about GNU/Linux, empowering users, was only possible of the existence of a standardized platform. Google's business model depends on disempowering users, which is why they work to maintain so much control over the platform of all their devices, and why they have put millions of arm devices in this world that cannot be easily repurposed as their users want. It is why I have a drawer of SBCs in my closet, all of which are functionally operational, but cannot be sustainably maintained or supported by the user... regardless of their experience or proficiency.

[oiaohm]

In addition Linux will not ever be able to create a standardized arm platform, if it were to lead, as we've seen the effects of Google's leadership on the arm platform.

This is wrong. The upstream Linux Kernel is pulling Google and others kicking and screaming into mainlining drivers and device tree for standardised arm platform.

You are missing how the upstream kernel is pulling them.

The mainline kernel does not list what is a security flaw this is intentional. Those not working with mainline Linux kernels will suffer security faults. So you have two choices work with upstream Linux kernel or have a defective product your choice.

Welcome to fun Google police on Android makers for drivers is more and more being written by the upstream kernel developers not Google.

Its been insanely slow process to get parties to understand they are writing core drivers they need the code peer reviewed or else they will have problems.

Google's business model depends on disempowering users, which is why they work to maintain so much control over the platform of all their devices, and why they have put millions of arm devices in this world that cannot be easily repurposed as their users want.

This is not new problem. This predates Google getting into the market. You think about your feature phones how many of those before android could you in fact flash a new OS into this number was almost zero. Horrible as it sounds google is less of a villain than its predecessors. Higher percentage of android devices can be repurposed than your prior windows mobile devices and before.

Google in a lot places are going to be pulled in by the linux kernel mainline gravity. Company like Apple who is making chip designs is going to be able to resist please note they need to make more income per product to do this.

The gravity of Linux kernel mainline is caused when you need to be cost effective as possible on prevent security flaws so you need to share stuff so it can be peer reviewed.

Something to be aware of out of all the different vendors who design soc chips for arm only 1 is not working on the mainline linux kernel up-streaming and that is Apple. This gravity well effect is not just happening because Google its also happen because of IOT and other groups where parties making devices are also after mainline linux kernel support to make their life simpler putting in security fixes.

Those in the arm space who ware not working in the direction of Linux kernel mainline who are not apple are having issues where their customers are asking for mainline Linux support and if they don't give it those customers are at times walking away with their money to a competitor.

So the idea that the Linux will not ever be able to create a standardized arm platform is missing what is currently happening. A 99 percent standardized arm platform is coming the central force is the Linux kernel.

Think about it once most arm chips support the mainline linux kernel with uboot it is simpler for parties like prism or pine64 to make their own vendor not locked devices. At this point the fight for userspace with google can start. Of course the first battle you have to win is the hardware vendors battle and that battle is almost complete won in the arm market 5 to 10 years to completely finish the process with the last holdouts other than apple.