Hacker News Reader

Comment by tcmart14 | Hacker News Reader

To be honest, Windows really should just rebase on to Linux. They can make their programs work on linux. But the OS isn't the future for Microsoft when it comes to profits. Their future is software and tooling. I don't follow MS's financials, but I bet their growth with Azure and Linux is far outpacing Windows license keys. I believe they have already indicated that Ubuntu is far more popular than their own server OS. The current and future cash cow is Microsoft making money off developer tooling/infrastructure and their apps.

And honestly, Microsoft isn't doing too bad at tooling. Had you told me 3 years ago I would be developing .NET applications, mostly ASP.Core, on linux, I woulda said you were stoned. But here I am today, cranking out .NET code that runs servers on Linux. And the tooling is pretty good and is just getting better.

View on HN

I don't think rebasing on to Linux offers them much. The current interoperability is "good enough", and the rest is UX - something Microsoft struggles with across the board. It's not for a lack of trying either, they've spent years now working on Fluent and it's still mediocre. Every time you think they make a fantastic decision, they turn around and add 10 more horrible UX choices (see Edge). It's so apparent it feels like as soon as the core design and dev team finalizes something, it gets sent to a penny pinching team who requires it have a bunch of unnecessary cheap bloat.

This is something that won't change on the shift to Linux. It has to be a deep cultural shift, until then the company is never going to prioritize experience.

> Windows really should just rebase on to Linux

No, it should not. The NT kernel architecture is drastically superior to the Linux monorepo.

Mind to elaborate a little more?

Drivers ABI, being micro-kernel based, even if everything is kind of bundled together, subsystems talk over in-kernel RPC (LPC), nowadays drivers and critical kernel subsystems are sandboxed on their micro-vms, hyper-v is a type 1 hypervisor (Windows runs on top of it when enabled, is just another guest), many kernel subsystems have long adopted C++ (hence VC++ /kernel since Vista, and Windows Implementation Library, a set of C++ templates for kernel, driver and userspace resource management), async IO since Windows NT early days, COM as OOP ABI (even if tooling is somehow clunky), subsystems just like mainframe OSes,....

Drivers with a stable API that don't break with each new minor version.

Why on earth would they do this and piss away their huge BC advantage?

Just open up to the WINE project. My understanding is, when you run a Windows 95 program on Windows 11, it essentially fires up an emulator anyways. WINE with the resources could very well offer the replacement at really no cost to Microsoft.

I'm assuming BC stands for backwards compatibility?

Yes.

> Their future is software and tooling. I don't follow MS's financials, but I bet their growth with Azure and Linux is far outpacing Windows license keys.

Azure outpaces everything else. Azure, under the covers, runs Hyper-V on Windows Server.

A massive number of VM's running on Azure are Linux, so investments in that area also pay cloud dividends.

Rebasing on Linux would be replacing the great part (the kernel), while keeping the broken part (shell, ads, millions of GUIs, …)

Reply on news.ycombinator.com