#dat

yoshuawuyts: what kind of schisms are there in node land these days?

mafintosh: https://github.com/datproject/dat/issues/827#is...

I'm not on twitter anymore, so in the dark :)

ralphtheninja: a few times core has broken something that didn't need fixing but people's complaint were heard

substack: aah, like readline in 8.1.something?

I haven't been following core that much lately, maybe i should now that I have more time on my hands, or maybe I shouldn't because I don't like politics

I've always been out on the fringes of userland

looks like I'll be speaking at nodeconf Argentina

pfrazee has mentioned Dat has content-addressable links a couple of times, any links that describe how that works?

karissa: wooooot! i was there last year. lots of fun

aldebrn: there is a paper

aldebrn: https://datproject.org/paper

ralphtheninja[m]: spend time here in userland instead of joining the bike shed

mafintosh: thanks, I will read it more carefully presently but quick question: I understood that (Rabin) chunks are content-addressed, is there any way to refer to a file by its hash in Dat?

aldebrn: the entire dat itself is content addressed

ralphtheninja[m]: quit twitter?

mafintosh: Wow, ok! I'll read carefully to understand. I was likely misled into a simplistic understanding because the same contents can get different Dat archive ids, but if I understand you, underneath that id is the content-addressed goodness.

aldebrn: the id of the dat is a public key signing the content

so its internally it is content addressed logs signed by a key pair

there is a static mode also, but we don't really use that

mafintosh: so peers in a Dat swarm are, quite like BitTorrent, asking for chunk hashes and getting the original chunks in exchange

aldebrn: yea exactly

except its live

so you can update the content

I ask all this because I was about to make a Dat blog article that said "this dataset is obtained by running the script with hash X on the original data with hash Y" and wanted to see if I could make X and Y more than just the text

It's not sufficient to refer to Dats hosting X and Y with their ids, because those could go away someday, even while the underlying data itself lived somewhere on the network.

is a "dat" synonymous with a "hyperdrive"? or is there other data around it?

noffle: yeah. there's a dat-storage module. Ir provides nice defaults for hyperdrive including key management in the home dir to prevent accidental key leakage

It*

I'll check that out

What happens if two persons try to update an hypercore/hyperdrive at the time? (they both have the private key)

karissa noffle also the peer discovery defaults, the remote storage, some extra metadata stuff for hyperdrives

dat is like a layer above hyperdrive that turns it into a user-facing set of tools and services

mafintosh: that's what I'm doing :)

noffle: yes but that was years ago

(benrogmans) @mafintosh just out of curiosity, what kind of consensus algorithm are you using for the multi-writer feature of the Dat protocol?

(benrogmans) And how's the progress ;) ?

@benrogmans it's eventual consistent

If two writable peers update a key st the same time the key has two values

Api wise get (key) always return an array

Up to you to pick the "best" one

If you put then the value overrides any existing ones deterministicly

@benrogmans i have one bug left and a bunch of docs :) had a very productive flight the other day

(benrogmans) Ok, so it is pretty straight forward

(benrogmans) That sounds promising :D

(benrogmans) Have you considered more complex algorithms like Paxos / Raft?

(benrogmans) I guess this article might answer my question: http://guide.couchdb.org/draft/consistency.html

cblgh: I guess that's what pfrazee talked about yesterday (AP/CP) ^

cblgh: lots of good plane hacking done btw :)

(benrogmans) @ralphtheninja oh wow, I missed the discussion above... Have to scroll up and catch up on the chat

mafintosh: do you have a command like a linter that detects if you have es6 features in the code?

pfrazee: no

pfrazee: i use some es6 sometimes

never babel

ok

and never promises

haha

:D

test

test2

mafintosh: niiice

ralphtheninja: ah yeah thanks

jhand: test ack

yay! think http://dat-chat.netlify.com/ is mostly working now

been trying to figure out why it won't let me set the domain to chat.datproject.org

nice, so that's a bot that's rebroadcasting chat events on a hypercore?

pfrazee: ya! so we have the hyperirc running on a server, collecting logs here. Then I have hypercored sharing it over websockets and it reads from that

very nice!

pfrazee: want to add the other channels we're tailing over hyperirc to that ui too!

jhand: yeah that'll be cool

jhand: http://chat.datproject.org/ works for me

ogd: ya just got it working. netlify wouldn't let me set the domain to chat.datproject.org because it "wasn't unique". so i set it to chat.datproject then changed it to the right one...

lol

gave a talk at bleeding edge austin last night and everybody said Dat was the coolest

woot

is that like straight edge

cause if those ppl like dat then we are def cool

yeah, but for web tech

nice so they only use gpl

exactly

it's a cool meetup, the organizer does a slideshow at the start showing what new features are in browsers since the last meetup

ohhh i see

yah

i would be interested in a meetup where the organizer did a summary of ways ppl have used blockchains to steal millions of dollars in the last month

lol!

obensource: i have an amplifier in my basement with your name on it

obensource: can i drop it off at your homestead some time?

ogd: aha!

ogd: yarp, but we just moved to the middle of nowhere to save money.

ogd: i'll have to come pick it up sometime

obensource: do u have goats

ogd: naw... I wish

ogd: with my folks in St. Helens

ogd: For up to a year to pay off debt, college etc, while we look for a place

ogd: have a garden tho! :D

obensource: nice. i believe parts of Twilight were filmed in st helens

ogd: they were! Actually in the lot next to the building where Liz/I got married!