Comment by theamk | Hacker News Reader

Comment by theamk | Hacker News Reader

Parent Full thread

Urbit “planet” is like a VM, but running an OS developed completely independently from existing computing words. There is a custom language, a custom network layer, a custom cryptography and custom identity system. There are even custom words for common terms. This has no GUI, but it does has a web interface that you can connect to.

From the practical standpoint, you have a linux process running from a data directory. You run it, and the you can access it via CLI and via web interface (which also has a terminal-like webapp). It then uses regular IP to talk to other urbits in P2P fashion.

From what I could find, the whole thing is full of magic and pretty opaque - you are only supposed to interact with it by running it and typing commands there. In particular, you’d probably want to back it up. The only instructions I could find is to stop it and tar it up, and back up manually. Better do it often if you want your “digital life” to be safe.

> custom cryptography

Just lost me.

> you are only supposed to interact with it by running it and typing commands

So, "the last computer you'll ever need" ... other than my phone, because the last time I wanted to type commands on a phone-sized device, was back when I was wet-dreaming over the Sharp Zaurus.

Given the artificial scarcity of the IDs and the semi-mystical woo describing the proposition, my spidey-sense is telling me some smart people have been busy devising an naivety-magnet, all the better to part people from their money ("buy a planet"...um, ok).

As far as I’m aware, there is no custom cryptography, and you can interact with it through a web UI now, not just through commands. Not sure where they got that idea.

Re custom crypto: random searches in github show plenty of it. I have no idea if this is an experiment or used in main code, but it is certainly present. Here is one example: [0]

> This raw interface is vaguely modeled on the haskell aos-signature package, but is written in terms of ed25519 primitives instead of general ECC and changes how linkage tags are computed...

Re command line: I was based it off the public docs. Go to "https://urbit.org", click on "install". The very end of the instructions say:

> Type (add 2 2) into the Dojo.

Maybe "install" is wrong section? I went to "docs" instead, from that "operators manual" seems most user-friendly. But it is text only as well:

> Let's join the ~dopzod/urbit-help channel. Use ctrl-x to switch from the Dojo prompt to the Chat prompt. Then:

> ~sampel-palnet:chat-cli/ ;join ~dopzod/urbit-help

The only reference to web UI I found was in "Using Bridge" section, but that seems a pre-requisite to set up the system, not part of the system itself.

If there are web UIs, you really need to document them better. Some screenshots would be great, for example.

[0] https://github.com/urbit/urbit/blob/7ce50ad75e23fcec0c8cb6ad...

Each planet can generate 2^32 moons, which are subsidiary processes. Your phone could run one of those.

You can also generate a comet, which is effectively identified by the hash of its public key. Those are free; while eventually planets have some notional advantages, notably a default positive reputation, at the moment there are no downsides.

Oh man I remember the first time my buddy showed me his Zaurus out front of a bar in NYC - he had some kind of Remote Desktop with a Linux shell in that tiny screen. He had to scroll around to see stuff with the stylus in a not-so-user-friendly a manner and the input was slow as balls but I was absolutely enamored. I at the time was very proud of my newly released Motorola Razr in my pocket yet here was my buddy with this magic device showing me up. Thanks for the trip :)

>There are even custom words for common terms.

I think this is like half the problem

Reply on news.ycombinator.com