#dat

(e-e-e) Creationix: that is an amazing tool. Thanks for the link.

@LzTjF5eAVC6xprPMGeNzqEP/EEBIGpM6hDaaY10hrz8=.ed25519:...the tools I already know, and that I expect will make life easy for others. Looking quickly at the ... http://wx.larpa.net:8807/%259a26CtDKg%2FNFcE6Lc...

creationix: let me take a look

lots of dat hacking happening at ccc atm!

oh nice

mafintosh: sounds fun. I'm glad you were able to go.

"lots of dat hacking happening at ccc atm!"<- 🤗

@mafintosh what are you working on?

lol dcrobins that emoji registered as a kanji for me

i was like "i wonder what that means"

lol

paste it into google translate and see the hug emoji

nice!

(rseger) @soni everyone having the link is how you support everyone being able to download/retrieve the files you post, yes. But they don't have to download/retrieve them from you. So you can whitelist trusted resources (or blacklist those who misbehave) on your personal machine/infrastructure without affecting the network at large's ability to serve the files to everyone.

(rseger)

(rseger) Of course, that assumes your feed is popular enough to have an "everyone" category of consumer but, if you're worried about DDoS, I'm assuming you've got that angle covered.

dcrobins: hacking on multiwriter! and then answering a bunch of dat questions

we are running a session as well in 30 min

YES moar multiwriter please

can I run dat behind cloudflare?

can I run dat behind websockets?

is websockets part of normal dat?

@mafintosh - awesome - hope it's fun! let me know how the session goes

idk anything about webrtc tho so idk if cloudflare supports webrtc

(rseger) to the best of my knowledge, dat doesn't run over websockets. I'm confident it could be extended to do so but that's a whole other can of worms.

(rseger)

(rseger) my gut says that cloudflare will mess with some of the networking tricks dat uses (hole punching, in particular). Enough so that I'd be shocked if you did manage to get it working behind a cloudflare deployment. Buuuut, I've obviously never tried. Probably the best way to figure it out is just to try?

here's the thing tho

you don't need hole punching when you have full control over port forwarding

everyone would be an incoming connection to the cloudflare servers, which would then proxy to my dat-based "seed" server

also it'd all be on port 80 or something

(rseger) massive change in topic,

(rseger) if I wanted to leverage the ability to only download specific files from a channel, where in the tech stack should I be looking?

(rseger)

(rseger) similarly, if I wanted to better leverage the versioning, and underlying merkle tree, where should I look first?

(rseger) that deployment should work fine

(rseger) cloudflare supports arbitrary tcp, yes?

I don't know

I'm not sure if the free tier does

(rseger) well, assuming it doesn't work at a higher layer (http, for example), you should be good to go. Though, I do remember seeing an http-dat somewhere. That might be worth looking into as well

I'm pretty sure the free tier includes shared address pools and stuff, and relies on HTTP "Host" header or SNI?

(rseger) sorry, I've no idea :(

(rseger) why are you so concerned about DoS attacks anyway?

rseger you can do {sparse: true} and then pick which files to download.

oh docs.datproject.org is down

jhand: o/ ?

rseger https://github.com/datproject/docs/blob/8c16c58...

competing game streamers are known for ddosing eachother to steal views

rseger you can also user hyperdrive and then do drive.download('file-i-want')

competing minecraft server owners are known for ddosing eachother to steal players

and so on

whoops, can see at https://datproject-docs.netlify.com/ for now

karissa: thanks, should be fixed/waiting for dns propagation

(rseger) thank you for the pointers! What I missed in that faq is, once you've declared a dat sparse, what's the best practice for monitoring the index? I stumbled into https://github.com/juliangruber/hypercore-index, am I barking up the right tree?

rseger what do you mean by monitoring the index?

dat can run over websockets and any arbitrary protocol if you write connection logic for it

WIP on dat-http: https://github.com/datproject/dat-http

(rseger) how best to know when there's a new file I might want to download

rseger good question...there are a variety of events on the hyperdrive object but i don't think new file is one of them

rseger you could probably listen to the replication stream on('data') event

rseger https://github.com/mafintosh/hyperdrive#var-str...

(rseger) @soni didn't realize gamers were so evil to each other. If you're looking specifically for tech to build a streaming platform on, you might look into some of the in-browser torrenting. Some of those are set up specifically to enable that kind of thing.

you wanna listen to the append event on the metadata feed

not streaming platform

just standalone streams

the best part of p2p is that if they ddos all peers all they're doing is harming themselves because they won't be getting any views from the saturated links

(rseger) thank you guys!

(rseger) @soni, this is what I was thinking of https://webtorrent.io/

yeah ik about webtorrent

it's just torrents

I'm looking for append-only streams

I think we both know something with that

(rseger) the tech supports streaming (that's what's going on with the movie in the upper right hand corner)

(rseger)

(rseger) I'm falling more and more in love with dat, don't get me wrong, it just sounds like webtorrent might be a tool more suited to your use case. Though, I admit, I'm still a little hazy about exactly what that is so I might be totally wrong.

livestreaming

it's a bit different from what webtorrent does

rseger there's also dat-shell you can use to cd into dat dirs, and cp files locally, even stream mp4 from a remote dat.

(rseger) ahh, well, the initial seeder will always be vulnerable to DoS in that instance, yeah. I'd love to hear if you can get (https://github.com/datproject/dat-http) working over cloudflare though!

(rseger) I am so using that to stream from my file server at home, thank you millette!

karissa, jhand you rebuilt docs? https://github.com/datproject/docs/issues/99 (Fixed now!)

millette: huh no just had made a DNS mistake and switched it back?

I've seen that bug before though

hmm, well js (ajax) links were broken yesterday but work now :-)

ya weird. last deploy was few weeks ago

Soni, yesterday, I suggested plugin two hypercores together via https://docs.datproject.org/hypercore#var-strea...

Thanks though, I'll keep an eye on it.

two independant keys, one feeding the other. That should isolate you a bit.

jhand, well thanks, whatever you did :-)