Hacker News Reader

Comment by SteveNuts | Hacker News Reader

This will be a very controversial prediction, but mark my words: Windows will eventually use the Linux kernel.

I strongly doubt they would ever do something that drastic. It offers few benefits and many roadblocks. It would be a monstrous amount of work, would throw into question many existing security-related certifications, break Microsoft's love of backwards-compatibility, etc.

Windows NT linage has supported UNIX since day one.

While the original support wasn't great, SUA was quite usable, until they decided to discontinue it on Windows Vista.

Nowadays we have WSL, which makes more sense, given how many folks buy Apple hardware and then complain UNIX isn't GNU/Linux.

> Windows NT linage has supported UNIX since day one.

yes, that's why the attempt to provide a Linux subsystem on top of the NT kernel (WSL1) was so successful they abandoned the approach entirely

WSL2 runs the full Linux kernel in a sidecar VM

Also known as, you don't know technically anything about Windows.

There is a very big difference between supporting UNIX, and Linux kernel syscalls ABI on top of pico processes, the technology from Drawbridge kernel taken out from Microsoft Research, which incidentally is also used to port MS SQL Server into GNU/Linux.

who cares? they still abandoned it

because it didn't work

You don't that is for sure.

As for the rest I could provide examples of how the BSDs and Solaris failed in similar attempts to clone Linux syscalls table, despite being UNIX, before Microsoft's attempt, but who cares?

> You don't that is for sure.

correct

> As for the rest I could provide examples of how the BSDs and Solaris failed in similar attempts to clone Linux syscalls table, despite being UNIX, before Microsoft's attempt, but who cares?

the BSD approach is still supported and part of FreeBSD, so presumably someone cares about that

whereas WSL1 is dead

I hope you’re right! They sort of did the same thing already with Edge given it’s just Chromium under the hood.

Reminds me of a specific thought experiment with a boat.

The Ship of Theseus https://en.m.wikipedia.org/wiki/Ship_of_Theseus

(For today’s 10000 (https://xkcd.com/1053/))

Why not just keep the existing kernel running with a small team? Drivers and backwards compatibility are critical to many of Microsoft's enterprise customers.

Now that isn't necessarily true for Windows running in the cloud. Drivers don't matter as much there.

I don’t disagree. How I think it will happen: Backwards compatibility will start to stray at Microsoft (their last bastion), leading to WINE becoming the go to tool for backwards compat support. Microsoft will create a linux variant or maybe just a DE and the rest will be history (just as Libre Office has begun to supersede MS Office). It is a matter of will and time that C-Suite will want to eliminate labor around Windows to maximize profits as it becomes more and more hardened in it’s feature set. Why maintain when you can utilize FOSS?

In what business segment is Libre Office superseding MS Office? Furthermore, NT supports "personalities". Why bother with a full Linux kernel when they could adopt the personality a la SUA?

Superseding in compatibility. I’m not talking about business segments.

They don’t have to adopt it, they will probably fork it.

> Superseding in compatibility.

It's going to be more compatible with Word than Word is?

Yes: https://news.ycombinator.com/item?id=38906331

It already does - SQL Server for Linux runs largely unmodified Windows userland on top of Linux kernel.

https://www.microsoft.com/en-us/sql-server/blog/2016/12/16/s...

Just before that: Microsoft will buy Canonical

Reply on news.ycombinator.com