Hacker News Reader

Comment by throwanem | Hacker News Reader

I doubt they'll attempt to disable execution of web content based on some "this is probably an emulator" heuristic. On the one hand, that'd instantly be circumvented into negligibility; on the other, the backlash would be epic.

I also severely doubt they'll attempt to prevent installation of application binaries from sources other than the Windows store. If anyone were going to be dumb enough to attempt such a move, it'd have been Apple, years before now, and they never even came close. Microsoft isn't run by morons, either.

View on HN

Webcontent is not a free platform. It does not even give you sockets which can interact with arbitrary network services without centralized control (no, websockets don't work because they can't listen, webrtc does not work because it needs rendezvous servers). And neither can it access your own filesystem. It is quite locked down. In a way you get compute cycles, but you don't get access to the machine's full capabilities.

And they could trivially forbid loading file:// URLs or disable features for those since few people use those anyway, that way you would lose persistence and would be dependent on something to even host your software.

I don't see where I made mention of any concept of freedom.

webrtc works fine. Rendezvous servers are a minimal cost, they're not involved beyond the setup, and you can run your own fairly easily.

I'd argue it works better than sockets, especially for gaming, since there's no need to mess around with NAT.

It's about freedom to run your own software, it is not cost or "works fine". Webrtc has a dependency on and is controlled by a 3rd party. You can't talk to someone in the next room via webrtc without some party running the server and providing a service to you. If they decide to lock you out you can't use sockets anymore.

With real sockets you can just send a multicast message to find all devices on the local network and talk to them directly.

You can use any rendezvous server in the world. Even without free ones, you could buy access for $1 per decade.

There is no third party in charge. There is just a requirement to use some party anywhere.

Imagine you're in china. Do you really want the basic ability for two devices to talk to each other depend on something that goes over the internet?

And they still have to agree on the same server, that provides a certain lock-in or inertia. Ultimately it means you have less control over your network connections than with real sockets.

And for the purpose of discussion this about the freedom of running software on your computer and webcontent not being equivalent in that regard. With webrtc signalling servers you're just kicking the ball down the road because then you need to be able to run a server with access to real sockets on some other machine, just to use those poor imitations of sockets.

> Do you really want the basic ability for two devices to talk to each other depend on something that goes over the internet?

Set up your hosts file or edit the js to point at a local server.

Of course, if you're in China you'd get in trouble making connections to an outside STUN server but if you're in China you're in trouble making connections to the outside anyways, and the STUN connection is probably the least of your worries.

> just to use those poor imitations of sockets

... that work across NATs without issues, as opposed to real sockets...

---

In any case, WebRTC doesn't strictly speaking need discovery servers at all. See http://blog.printf.net/articles/2013/05/17/webrtc-without-a-... It still uses STUN for public-ip discovery, but comments indicate that can be bypassed by just configuring the connection to use a local IP for STUN. Unsurprisingly, not many people seem to actually care about that.

> that work across NATs without issues

You say that as if stun didn't exist in the udp/tcp realm. bittorrent clients in fact engage in serverless stun by using mutually reachable peers. real sockets are strictly more powerful than webrtc, period. Also, that argument is becoming more and more obsolete as ipv6 adoption increases.

And again, this is not about utility, it's about control and freedom.

> In any case, WebRTC doesn't strictly speaking need discovery servers at all.

Even if you don't use a server you need to exchange one-time connection tokens out of band. You can't re-establish connections at a later point either. And it does not support multicast either. So no, you can't just have devices talk to each other.

Anyway, you're not even responding to my central argument. Web content is not a substitute for unfettered native access because you need native access to things (e.g. running a server) to even be able to run web content.

I love that you consider trivial the extrapolation from "Microsoft bans emulators in Xbone and Windows Store" to "no one can run nginx without authorization".

On windows store, not just on xbox. The next steps need no imagination, they already happened on other platforms, see the iphone or windows RT. For all intents and purposes they are general purpose computing devices but everything needs to be approved by a single vendor.

> "no one can run nginx without authorization".

Your freedom does not end when the last person on earth who could run nginx cannot do so any longer. It ceases when you can't do it on the device you own on a whim of a 3rd party.

Yes, and if you're so monomaniacal about freedom as to utterly neglect everything else, that's a serious concern.

If people are willing to accept restrictions on what can be obtained or sold on the store (falsely in that name of "safety"), what makes you think they wont accept restrictions on what websites they are allowed to load, or more likely, what website owners are allowed to host?

I don't know. Nuance?

Reply on news.ycombinator.com