Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

The Linux kernel hidden inside Windows 10

github.com

Post reply on HN
Print view View original

31–40 of 166 posts

Re: The Linux kernel hidden inside Windows 10

#31

Post by sdrinf »

Zardoz84 wrote: post #28
sdrinf wrote: post #23

Earlier quoted context omitted.

Screenshot: http://imgur.com/a/5Fcr9 Short summary: * Ubuntu, and all children apps run under current user's credentials, but are launched under an Lxssmanager servicehost process (as opposed to explorer.exe) * ProcMon can interact (send sigterm / etc to) unix processes; but as shown on htop, unix processes are not aware of windows ones; nor have I found any ways to send signals backwards * API compatibility is excel…

And there is some strange behaviour with the keys. Actually using LXSS become far lees powerfull to the users of Conemu thanks to this. Vim and NeoVim does weird things with the arrows, on mc I can't navigate on the filesytem with the cursors, and htop does weird things.

I suspect that might be conemu, and compatibility thereof. Specifically, running the bog-standard bash prompt (with conhost.exe) makes this a no-repro; all keys / arrows works in MC & htop

Re: The Linux kernel hidden inside Windows 10

#32

Post by wangchow »

Amezarak wrote: post #29
satysin wrote: post #27

Earlier quoted context omitted.

Indeed, and the rumours are that Project Astoria worked too well which scared Microsoft. If Android apps "just worked" on Windows then why would developers bother using their new Universal Windows Platform.

Which is really unfortunate. I can't blame Microsoft for it, but the app-gap between WP/WinMo 10 and Android is getting untenable for me. I'm thinking about switching back, not because I think Android is better (I don't think it is at all), but simply because I'm starting to feel left out when all my family and friends are using, e.g., SnapChat to keep in touch and I...can't. It particularly riles me up that there we…

Honestly I use the app-gap as a feature. Not having the fluff is great because snapchat is making people illiterate. We went from real words to abbreviations to no words and minimal attention span.

On the other hand now that Uber and bank of america have a universal app it feels more complete.

Re: The Linux kernel hidden inside Windows 10

#33

Post by rrdharan »

sdrinf wrote: post #23
coverband wrote: post #8

I'd love to see a video of the demo session for this; really curious how ProcMon and AV tools will deal with ELF binaries in this release.

Screenshot: http://imgur.com/a/5Fcr9 Short summary: * Ubuntu, and all children apps run under current user's credentials, but are launched under an Lxssmanager servicehost process (as opposed to explorer.exe) * ProcMon can interact (send sigterm / etc to) unix processes; but as shown on htop, unix processes are not aware of windows ones; nor have I found any ways to send signals backwards * API compatibility is excel…

> unix processes are not aware of windows ones; nor have I found any ways to send signals backwards

I haven't tried it out, but this project claims to offer support for doing exactly that:

https://github.com/xilun/cbwin/

Re: The Linux kernel hidden inside Windows 10

#34

Post by Mahn »

satysin wrote: post #27
vvanders wrote: post #7

Fun tidbit is it seems the original reason behind this was Android support in WinPhone10 but when that was axed they migrated it to Ubuntu on desktop. If true that's a pretty nifty pivot.

Indeed, and the rumours are that Project Astoria worked too well which scared Microsoft. If Android apps "just worked" on Windows then why would developers bother using their new Universal Windows Platform.

That's a shame. I mean, I get the concern, but full Android compatibility would have done wonders to adoption, which is what their phone platform desperately needs. They could have always lured developers later with a substantially better UWP on top of it.

Re: The Linux kernel hidden inside Windows 10

#35

Post by Zekio »

Zardoz84 wrote: post #28
sdrinf wrote: post #23

Earlier quoted context omitted.

Screenshot: http://imgur.com/a/5Fcr9 Short summary: * Ubuntu, and all children apps run under current user's credentials, but are launched under an Lxssmanager servicehost process (as opposed to explorer.exe) * ProcMon can interact (send sigterm / etc to) unix processes; but as shown on htop, unix processes are not aware of windows ones; nor have I found any ways to send signals backwards * API compatibility is excel…

And there is some strange behaviour with the keys. Actually using LXSS become far lees powerfull to the users of Conemu thanks to this. Vim and NeoVim does weird things with the arrows, on mc I can't navigate on the filesytem with the cursors, and htop does weird things.

you can get ubuntu bash into ConEmu with just:

-new_console bash ~

Re: The Linux kernel hidden inside Windows 10

#36

Post by __jal »

vram22 wrote: post #21

Microsoft has actually been doing Unix related stuff from way back - a fact that might not be known to some. Xenix, then SFU (Services For Unix - for interop - on Win NT or 2K), plus they had (part of?) a POSIX subsystem around then or earlier- I used it a bit for C utility dev work on WinNT), etc.

They didn't build SFU - the first thing called that was a rebranded MKS toolkit, from Mortice Kern Systems (later, I think, bought by somebody else). The second thing called named SFU was a different POSIX layer built by some company I can't remember the name of.

Remembering having to use it (the MKS version) to port some Solaris software, I recall my team mates having an alternate expansion for the acronym that isn't that hard to guess.

Re: The Linux kernel hidden inside Windows 10

#37

Post by sdrinf »

rrdharan wrote: post #33
sdrinf wrote: post #23

Earlier quoted context omitted.

Screenshot: http://imgur.com/a/5Fcr9 Short summary: * Ubuntu, and all children apps run under current user's credentials, but are launched under an Lxssmanager servicehost process (as opposed to explorer.exe) * ProcMon can interact (send sigterm / etc to) unix processes; but as shown on htop, unix processes are not aware of windows ones; nor have I found any ways to send signals backwards * API compatibility is excel…

> unix processes are not aware of windows ones; nor have I found any ways to send signals backwards I haven't tried it out, but this project claims to offer support for doing exactly that: https://github.com/xilun/cbwin/

| Note that outbash.exe listens on 127.0.0.1, but validates that processes that establish a connection are running as the same user as outbash.exe (this check is implemented starting with version v0.7). Therefore, it can be used on multi-user computers.

(from: https://github.com/xilun/cbwin/ )

So, I suspect this uses TCP & networking to do that; which does work: you can listen to a port in LXSS, and that will be accessible via 127.0.0.1 ;and you can connect to 127.0.0.1 from LXSS, and it will be routed to eg. windows listeners . Above was specifically about process-based signals, and non-network-hacks.

Re: The Linux kernel hidden inside Windows 10

#38

Post by kragen »

whyenot wrote: post #25
vram22 wrote: post #21

Microsoft has actually been doing Unix related stuff from way back - a fact that might not be known to some. Xenix, then SFU (Services For Unix - for interop - on Win NT or 2K), plus they had (part of?) a POSIX subsystem around then or earlier- I used it a bit for C utility dev work on WinNT), etc.

Of course Microsoft then licensed XENIX to the Santa Cruz Operation which after several years renamed it to SCO Unix. SCO eventually split apart, but the piece that ended up with SCO Unix renamed themselves the SCO Group. The SCO Group with (possibly indirect) financial backing from Microsoft, attempted to use XENIX in a number of copyright infringement lawsuits to attack Linux vendors.

Hmm, there are a variety of inaccuracies in your short statement. Please carefully read the below so that you can be protected from propagating these errors in the future.

· SCO Unix, although it incorporated some Xenix code, was very different from Xenix; it derived a lot from SVR4.

· SCO basically went out of business, because their value proposition was "Unix, but on a regular PC so you don't have to buy an expensive RISC workstation." This stopped being a useful value proposition around 1997 because they couldn't keep up with Linux. So they sold the SCO Unix product line to Caldera, a Linux distributor, in 2001. I'm not 100% sure but I don't think any employees moved from SCO in Santa Cruz, California, to Caldera, which was in Utah. Maybe someone stuck around to smooth the transition?

· Microsoft's backing of SCO was direct; they "bought a license" in 2003. Maybe they also did some indirect backing that I don't know about.

· Caldera basically failed in the Linux distribution space in 2002, and the investors booted out CEO Ransom Love and replaced him with Darl McBride, formerly of Novell, IKON, a couple of startups, and Franklin Covey (!).

· The copyright infringement lawsuits weren't based on Xenix. Nobody claimed Linux had copied from Xenix. Rather, they were based on Bell Labs Unix, from the 1970s, and AT&T Unix System Labs System V Unix. SCO had supposedly acquired the copyrights to these sometime in the 1990s, and in fact them giving permission is how the Lions book was legally republished in 1996 or 1997, but it turned out that in fact what they had acquired was not the copyright ownership, but a sublicenseable license to the copyright. The actual copyright rested with some company Novell had bought up at some point, so the lawsuit got thrown out of court.

· The SCO Group sued not only Linux vendors, notably IBM, but also Linux users, notably AutoZone and DaimlerChrysler. This is an error of omission, but it's crucially important.

Re: The Linux kernel hidden inside Windows 10

#39

Post by GauntletWizard »

Amezarak wrote: post #29
satysin wrote: post #27

Earlier quoted context omitted.

Indeed, and the rumours are that Project Astoria worked too well which scared Microsoft. If Android apps "just worked" on Windows then why would developers bother using their new Universal Windows Platform.

Which is really unfortunate. I can't blame Microsoft for it, but the app-gap between WP/WinMo 10 and Android is getting untenable for me. I'm thinking about switching back, not because I think Android is better (I don't think it is at all), but simply because I'm starting to feel left out when all my family and friends are using, e.g., SnapChat to keep in touch and I...can't. It particularly riles me up that there we…

There's a couple really good reasons why they can't do that, trademark, PR and user trust.

They need to defend their trademark, or it will be genericized. Not that that's a big deal, but it can be a thing.

More importantly, if the windows store is filled with crappy snapchat apps, it hurts their brand. If they release an update that breaks those apps, it hurts their brand. It's a no-win situation. Then there's hacks of third party apps, that are spun as being Snapchat's problem: http://www.cnn.com/2014/01/01/tech/social-media/snapchat-hac...

Snapchat is in a particularly bad position for the last bit, because what they're selling is fundamentally a lie, and having that lie exposed hurt them a lot.

Post reply on HN