limbo_ has quit [Read error: Connection reset by peer]
infinity0 has quit [Ping timeout: 240 seconds]
infinity0 has joined #ipfs
limbo_ has joined #ipfs
jamiew has quit [Quit: zzz]
jaboja has quit [Ping timeout: 246 seconds]
arpu has quit [Ping timeout: 260 seconds]
arpu has joined #ipfs
chungy has quit [Ping timeout: 255 seconds]
scde has joined #ipfs
chungy has joined #ipfs
jaboja has joined #ipfs
ygrek has quit [Ping timeout: 248 seconds]
aboodman has quit [Quit: aboodman]
aboodman has joined #ipfs
erictapen has quit [Ping timeout: 258 seconds]
jokke has joined #ipfs
<jokke>
hi
<jokke>
i just started out with ipfs
<crankylinuxuser>
hello
<jokke>
i set up the daemon on my laptop and on my server
<crankylinuxuser>
cool stuff, aint it ? :)
<jokke>
i want to use my server as a public gateway
<jokke>
definately!
<crankylinuxuser>
oops, waitasec
<jokke>
but experience a weird problem: the response is very slow when proxied through nginx
<jokke>
i don't understand why
<jokke>
if i run curl localhost:8080/ipfs/... on the server it's instantly there
<jokke>
but if i use my domain name it takes forever to respond
<jokke>
i use the nginx for a lot of stuff
<jokke>
so i doubt it's that
<crankylinuxuser>
ok im back
<jokke>
wb
<crankylinuxuser>
Yeah, Im a squid and apache person. Havent had much experience with nginx
<crankylinuxuser>
but about the public gateway, as long as you know what you're getting into, with DMCA or relevant copyright laws, then you'll be fine.
<jokke>
good point...
<crankylinuxuser>
and also, be aware that there are some unscrupulous people who've tried trading "unsavory" things over IPFS. The relevant hashes were banned. Also, it's public knowledge whom asks for what hash.
<crankylinuxuser>
It's a heavy use of encryption, but there is no privacy.
<crankylinuxuser>
For example, do you have a IPFS share I could request that you put in the network?
anewuser has quit [Ping timeout: 260 seconds]
galois_d_ has joined #ipfs
<crankylinuxuser>
jokke, do you have a share I could show an example/
anewuser has joined #ipfs
galois_dmz has quit [Ping timeout: 276 seconds]
<mayli>
> The relevant hashes were banned - you can ban hashes in ipfs?
<whyrusleeping>
no
<whyrusleeping>
hashes can be added to a list of hashes that the ipfs.io gateway will refuse to serve
aboodman has quit [Quit: aboodman]
<whyrusleeping>
crankylinuxuser: make sure to qualify properly when you make statements like that
<crankylinuxuser>
my apologies
crankylinuxuser is now known as cranky-away
zuck05 has quit [Remote host closed the connection]
zuck05 has joined #ipfs
scde has quit [Quit: Leaving]
tulior has quit [Ping timeout: 255 seconds]
o33 has joined #ipfs
o33 has quit [Client Quit]
_whitelogger has joined #ipfs
<jokke>
whyrusleeping: any ideas why my gateway might be so solw?
<jokke>
*slow
<jokke>
it's way slower than ipfs.io at least
<whyrusleeping>
jokke: hrm... and you said its much faster if you go through localhost while sshed in?
<jokke>
yeah
<jokke>
_much_
<whyrusleeping>
huh
<whyrusleeping>
i really dont know, my first guess would be weird nginx config stuff
<whyrusleeping>
lgierth handles all that for us
<jokke>
hm
<jokke>
the slowness is only for my ipfs gateway
<jokke>
i proxy alot of stuff through nginx and never had such a problem
<jokke>
do i need to proxy anything else than 8080?
<whyrusleeping>
no, you shouldnt have to
<jokke>
hm
<jokke>
it's not my connection either since if i use the domain name on the server (instead of localhost:8080) it's just as slow
<jokke>
and not just once, which would be understandable, but subsequent requests slow as well
<jokke>
*are slow
<whyrusleeping>
ooooo
<whyrusleeping>
do you have dns caching set up on your server?
<jokke>
nginx caching?
<whyrusleeping>
no, just dns
<jokke>
oh
<jokke>
hm
<jokke>
how can i find out? :D
<whyrusleeping>
whats in your /etc/resolv.conf ?
<jokke>
This file is managed by man:systemd-resolved(8). Do not edit
<whyrusleeping>
yeah, so whats going on here is that the ipfs gateway has this neat feature
<whyrusleeping>
where if you access is via a domain name
<whyrusleeping>
it checks if the domain name has a text record
<whyrusleeping>
and will serve something from that text record at the root
<whyrusleeping>
(this is how ipfs.io leads to a webpage)
<whyrusleeping>
so the speed difference is ipfs making that dns request
<whyrusleeping>
which would be near instantaneous if your server were properly caching dns queries
<abbiya>
ipfs is comsuming about 200 mb ram on that mobile
jonnycrunch has quit [Quit: leaving]
jonnycrunch has joined #ipfs
jonnycrunch has quit [Client Quit]
jonnycrunch has joined #ipfs
<abbiya>
whyrusleeping: what can i do with a device(have python3, go, git, curl, tar, arm 32) like this, suggestions ?
<whyrusleeping>
abbiya: hrm... make an app that lets you share pictures to people nearby through ipfs
<whyrusleeping>
you could use nfc to transfer a hash
<whyrusleeping>
then request the data via ipfs on the local wifi
<abbiya>
cool
<whyrusleeping>
i'm fascinated by mobile applications that allow me to deal directly with the people around me
<whyrusleeping>
would be really awesome to have an instagram/twitter type feed of things posted by people who are physically close to you
jonnycrunch has quit [Quit: leaving]
jonnycrunch has joined #ipfs
jhand has quit [Quit: Connection closed for inactivity]
jonnycrunch has quit [Client Quit]
jonnycrunch has joined #ipfs
<abbiya>
thank you. will do something about this.
jonnycrunch has quit [Client Quit]
jonnycrunch has joined #ipfs
aboodman has quit [Quit: aboodman]
jonnycrunch has quit [Client Quit]
aboodman has joined #ipfs
jonnycrunch has joined #ipfs
jonnycrunch has quit [Client Quit]
jonnycrunch has joined #ipfs
<abbiya>
*nearby, how ipfs connects to near by nodes, do i start without any bootstrp nodes ?
<whyrusleeping>
abbiya: ipfs uses mdns to find and connect to nearby peers
<tomatopeel>
ChrisMatthieu: really cool, one question that comes to mind watching your broadcast, the part where a computation request comes in and it's just some trivial calculation I think: surely the computational requirements of communicating a computation across the network and then receiving the answer back are quite expensive...
<tomatopeel>
So I guess I'm wondering about any guidance/guidelines on how to know when distributed computation is the approach to take
<tomatopeel>
first thing that comes to mind is proof of work
<tomatopeel>
and obviously there can be other benefits like HA and all that... It'd be nice to have some idea though of when it might be time to think about distribution of actual computations
<abbiya>
can multiple nodes add data to a common directory ?
<tomatopeel>
just need to measure how much computational power every user inputs, and then when some coin gets mined, distribute the value evenly across everybody who contributed...
dconroy has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
sprint-helper1 has joined #ipfs
m0ns00n has quit [Quit: quit]
mvollrath has quit [Ping timeout: 246 seconds]
rflot_ has joined #ipfs
iinaj_ has joined #ipfs
jbenet_ has joined #ipfs
dmj`_ has joined #ipfs
daviddias_ has joined #ipfs
zmanian_____ has joined #ipfs
AphelionZ__ has joined #ipfs
hosh_ has joined #ipfs
ehd_ has joined #ipfs
elliottcable_ has joined #ipfs
zuck05 has quit [Remote host closed the connection]
kumavis_ has joined #ipfs
dvn- has joined #ipfs
zuck05 has joined #ipfs
robogoat has quit [Ping timeout: 248 seconds]
conway has joined #ipfs
Guest89066 has quit [Ping timeout: 260 seconds]
dmj` has quit [Ping timeout: 246 seconds]
s4y has quit [Ping timeout: 246 seconds]
daviddias has quit [Ping timeout: 246 seconds]
jbenet has quit [Ping timeout: 246 seconds]
hosh has quit [Ping timeout: 246 seconds]
zmanian____ has quit [Ping timeout: 246 seconds]
ehd has quit [Ping timeout: 246 seconds]
iinaj has quit [Read error: Connection reset by peer]
kumavis has quit [Ping timeout: 246 seconds]
dvn has quit [Ping timeout: 246 seconds]
AphelionZ_ has quit [Read error: Connection reset by peer]
elliottcable has quit [Read error: Connection reset by peer]
voldyman has quit [Ping timeout: 246 seconds]
rflot has quit [Ping timeout: 246 seconds]
aristid2 has quit [Ping timeout: 246 seconds]
cranky-away has quit [Ping timeout: 246 seconds]
daviddias_ is now known as daviddias
dmj`_ is now known as dmj`
voldyman_ has joined #ipfs
jbenet_ is now known as jbenet
sprint-helper has quit [Read error: Connection reset by peer]
iinaj_ is now known as iinaj
aristid2_ has joined #ipfs
hosh_ is now known as hosh
ehd_ is now known as ehd
kumavis_ is now known as kumavis
rflot_ is now known as rflot
elliottcable_ is now known as elliottcable
s4y_ has joined #ipfs
s4y_ is now known as s4y
robogoat has joined #ipfs
corby has quit [Ping timeout: 246 seconds]
mvollrath has joined #ipfs
Guest75836 has joined #ipfs
zuck05 has quit [Remote host closed the connection]
zuck05 has joined #ipfs
robogoat has quit [Ping timeout: 240 seconds]
robogoat has joined #ipfs
spossiba has quit [Ping timeout: 240 seconds]
spossiba has joined #ipfs
zuck05 has quit [Remote host closed the connection]
zuck05 has joined #ipfs
m0ns00n has joined #ipfs
elima has quit [Remote host closed the connection]
zuck05^^ has joined #ipfs
corby has joined #ipfs
zuck05 has quit [Read error: Connection reset by peer]
zuck05^^ has quit [Remote host closed the connection]
bielewelt has quit [Quit: Leaving.]
m0ns00n_ has joined #ipfs
mildred4 has joined #ipfs
mildred has joined #ipfs
mildred3 has quit [Ping timeout: 255 seconds]
mildred2 has quit [Ping timeout: 260 seconds]
neuthral has joined #ipfs
homs has joined #ipfs
<homs>
hi. i have a question. why does google have in their search links to content I only shared between 3 of my machines?
Caterpillar has joined #ipfs
m0ns00n_ is now known as Guest36324
jamiew has joined #ipfs
ashark has joined #ipfs
<homs>
any ideas?
<neuthral>
google chrome?
<homs>
nope, i use firefox with sane plugins.
<homs>
noscript, priv badger, ublock
erictapen has quit [Ping timeout: 248 seconds]
<homs>
and one link was even on a private tor site. So a different question.... how would I watch all the hashes being transmitted if I wanted to build a IPFS search engine?
<Kubuxu>
homs: when you are typing in URL most browsers will also run Google Search on it
<Kubuxu>
so ..
<homs>
yeah but privacy badger keeps firefox from doing it
rodolf0 has joined #ipfs
robattila256 has quit [Ping timeout: 240 seconds]
<homs>
ive watched that with wireshark. no data until i initiate a search
<Kubuxu>
you could get info about good portion of content in IPFS by listening to the DHT announcements
<homs>
that sounds like the same way of finding tor hidden sites.
pat36 has joined #ipfs
<homs>
all I know is, that content I didnt share ended up being indexed by google. If I cant stop it, I'd like to be able to build my own index.
<homs>
^^ share, outside my 3 computers
<homs>
so listening to the DHT is a good way? Is there a ipfs command that gives me a continous stream of DHT announcements?
<homs>
The obvious is, "dont use IPFS for sensitive data". Yeah, I can use Tox or encrypted dropbox for secret stuff. But I wasnt expecting my semi-private hashes to end up on google so damned quickly.
<jcgruenhage>
I'd say using IPFS for sensitive data is totally fine, just encrypt it before :D
aseriousgogetta has quit [Ping timeout: 258 seconds]
aseriousgogetta has joined #ipfs
dconroy has joined #ipfs
<homs>
yeah, but the surprise was "why are my semiprivate hashes being indexed by google?"
<homs>
hmm. that seems like a good place to start. Especially if google's doing this, I'd expect that they probably have a dozen or few hundred machines dedicated to listening and requesting.
zuck05 has joined #ipfs
ulrichard has quit [Remote host closed the connection]
Guest36324 has quit [Quit: quit]
fzzzr has joined #ipfs
m0ns00n has quit [Quit: quit]
fzzzr has quit [Client Quit]
<ChrisMatthieu>
@tomatopeel You will need to write distributed apps slightly different from local apps. Short computations do get expensive due to network traffic. I find that if your computations run 10+ seconds or do not require immediate responses then network latency is no longer an issue.
<ChrisMatthieu>
@tomatopeel ^
<ChrisMatthieu>
You could absolutely use computes.io for bitcoin mining :)
erictapen has joined #ipfs
zuck05^^^ has joined #ipfs
spossiba has quit [Ping timeout: 248 seconds]
spossiba has joined #ipfs
mildred has quit [Ping timeout: 248 seconds]
zuck05 has quit [Ping timeout: 240 seconds]
zuck05^^^ has quit [Remote host closed the connection]
zuck05 has joined #ipfs
zuck05 has quit [Remote host closed the connection]
zuck05 has joined #ipfs
zuck05 has quit [Remote host closed the connection]
zuck05 has joined #ipfs
zuck05 has quit [Remote host closed the connection]
stavros has joined #ipfs
<stavros>
hello
<stavros>
does anyone have a good resource on how directories are represented on ipfs? maybe there's a single example file with the internal representation?
<stavros>
i'm mainly wondering this: say i have a static website, i publish it, and then i change a file on it and publish it again. will the already-published files retain their hashes? is the directory just a tree of pointers, and so oldhash/file1.txt can be the same as newhash/file1.txt?
<homs>
the hash of that file will stay consistent. but if the graph it points to changes then the overall hash changes.
<homs>
for example, you have 100MB of fimages and 5M of text. You change the text. The hashes for the images are the same and will point at the same blocks underneath. The text hashes change, and the overall hash changes.
<stavros>
homs, by "the graph it points to" you mean for a parent directory?
<stavros>
right, that's what i would expect
<stavros>
so if i pin the new overall hash, only the changed files need to be fetched, correct?
zuck05 has joined #ipfs
Caterpillar has quit [Quit: You were not made to live as brutes, but to follow virtue and knowledge.]
zuck05 has quit [Remote host closed the connection]
zuck05 has joined #ipfs
stavros has quit [Ping timeout: 268 seconds]
Rogu3Nod3[m] has left #ipfs ["User left"]
tilgovi has joined #ipfs
droman has joined #ipfs
tilgovi has quit [Remote host closed the connection]
tilgovi has joined #ipfs
bitbit has joined #ipfs
<bitbit>
Hi! is there a way with ipfs-js to connect between peers when they are near each other, even if the router in the office is down, etc?
dtrue has joined #ipfs
[X-Scale] has joined #ipfs
toXel has quit [Quit: Ping timeout (120 seconds)]
<Mateon1>
"If the router is down", No, maybe if bluetooth transport ever gets implemented. In cases without internet access (but with connection between computers) ipfs can use mdns, but I'm not sure if the Javascript version has that
toXel has joined #ipfs
tilgovi has quit [Remote host closed the connection]
X-Scale has quit [Ping timeout: 255 seconds]
[X-Scale] is now known as X-Scale
<bitbit>
so what is the connection between comupters that you're refering to?
<Mateon1>
bitbit: A router or a switch that is not connected to the internet will work
<SchrodingersScat>
Mateon1: ham radio
<Mateon1>
SchrodingersScat: If it can speak ethernet packets, it will work
<Mateon1>
SchrodingersScat: Although, isn't encryption over ham radio banned?
<SchrodingersScat>
Mateon1: I'm assuming a post-apocalyptic zombie situation where the FCC or /whoever/ cares about that would be disbanded by then.
<bitbit>
lol
dconroy has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
jungly has quit [Remote host closed the connection]
<yuvipanda>
I'm going to write a small blog post about installing it, what you can use it for now, and future plans :D
espadrine has quit [Ping timeout: 268 seconds]
rodolf0 has quit [Ping timeout: 246 seconds]
Caterpillar has joined #ipfs
erictapen has quit [Ping timeout: 255 seconds]
chungy has joined #ipfs
<xelra>
If I have just a single node connected to the internet and I have a thousand nodes on my intranet, would everyone still be able to discover the entirety of ipfs through that one node?
Encrypt has quit [Quit: Quit]
<SchrodingersScat>
xelra: good question
<xelra>
It's a question that came up in a conversation, with a friend who works in the space industry, about future drones on Mars and whehter ipfs is a viable solution to have them connected. Especially when there's a mothership that hosts multiple drones.
<SchrodingersScat>
xelra: I'm also concerned about mars.
aboodman has joined #ipfs
<SchrodingersScat>
xelra: because presumably you don't want every node on mars DHT'ing with every node on Earth1.0
rodolf0 has joined #ipfs
<xelra>
Yeah, I was really advertising it, told him that it's a system that was specifically designed for data transfer between planets, but at the first serious question I had to throw the "dunno" towel. :/
<whyrusleeping>
the DHT is just a tool ipfs uses for now
<whyrusleeping>
it can be easily swapped out for a different system
<whyrusleeping>
or made to be smarter about locality
<Kythyria[m]>
Like circuit-switching? :P
<SchrodingersScat>
whyrusleeping: i think this relates to previous talks. like can we have a 'bridged' mode where they understand they're the gateway to other nodes?
* Kythyria[m]
is still frowny at proposals to replace IP with something equivalent to running a miniscule profile of HTTP and every router is a proxy.
<SchrodingersScat>
the last IPv4 address was assigned in 1994
<Kythyria[m]>
IPv6, not just IPv4
m0ns00n has joined #ipfs
<whyrusleeping>
xelra: think about it this way, people using ipfs on mars don't need to talk to earth to find content that is also on mars
<whyrusleeping>
if the content doesnt exist on mars, they could send their wantlists (very slowly) to earth, and then have an earth node (very slowly) send them the content
<xelra>
Yes, that part is clear. But what if they want to get content that is only on nodes on Earth?
<whyrusleeping>
after that, the next mars node that requests that content doesnt need to fetch it from earth
<robertomurta[m]>
I think the main idea is not replace IP, but build a distributed network over any protocol available
<xelra>
Yeah, that.
<xelra>
But who relays that? They're not connected to earth.
<whyrusleeping>
for that, you could have a bridge node
<whyrusleeping>
that you send your wantlists to, and then they do communication with earth
<robertomurta[m]>
ipfs could run over cjdns, wich is distributed and uses ipv6 private address
<xelra>
Exactly. Bridge node. Is that hypothetical or does it exist?
<whyrusleeping>
hypothetical for now, shouldnt be too hard to implement
<xelra>
In reality I'd expect some laser-relay-station to handle the bridging and making ipfs available to the rest of Mars.
<SchrodingersScat>
heh, you kids love your lasers
<xelra>
Of course that introduces the problem of netsplits, doesn't it?
<Kythyria[m]>
robertomurta: At which point it's running over an overlay network. Yo dawg I heard you like overlay networks...
espadrine has joined #ipfs
bnwilson has joined #ipfs
<xelra>
Well, I guess I'll advertise it again at the next table-top-game session. For the motherprobe. It's supposed to land a main spacecraft which then sends out drones in multiple directions that collect data. I told him that ipfs might be ideal for that, because every drone would just be a node and all the info will just be available on the mother-drone. Something flawed in that argumentation?
<Kubuxu>
Someone here has Cygwin running?
<xelra>
Kubuxu: I do.
<Kubuxu>
xelra: can you run `uname -s` in Cygwin?
<xelra>
CYGWIN_NT-10.0
<Kubuxu>
xelra: and can you check `uname -o`
<xelra>
Cygwin
jamiew has quit [Quit: bye]
m0ns00n is now known as Guest36324
<tomatopeel>
whyrusleeping: why not very effective for bitcoin re: fully distributed mining pools, out of interest? Is it just that the diff between regular computers' clock cycles vs. the big boy miners G/C-PU's is ridiculous huge, or some more interesting reason?
<whyrusleeping>
tomatopeel: because a single bitcoin asic is comparable in hashrate to a very large number of normal computers
<whyrusleeping>
the total potential hashrate of all the browsers of all the worlds computers is quite small compared to the entire bitcoin network
<xelra>
Isn't distributed mining pools though what people do for Bitcoin mining? Even with ASICs?
rodolf0 has quit [Read error: Connection reset by peer]
<tomatopeel>
ah ty
* SchrodingersScat
puts on his big boy miner pants
<whyrusleeping>
xelra: yeah, they do
<SchrodingersScat>
xelra: likely, although if they have enough ASICs they could pool them together into their own network, if they have enough to actually solve a block.
<whyrusleeping>
though i guess what you would have to figure out is using an asic from the browser
<SchrodingersScat>
not sure what you mean, normally the mining client has a url:port and a username/password
<whyrusleeping>
SchrodingersScat: but thats for centralized mining pools
<jokke>
If i have the public key of a peer, can i validate that an ipfs hash originates from that peer?
<xelra>
I wonder whether it will be the same for filecoin. That I have to be part of a pool, just so that no one else is serving the request first.
<aboodman>
jokke: i’m not sure, but on the surface that doesn’t make sense to me. what if two different peers both added the same content. they’d have the same hash.
<aboodman>
what does it mean to “originate” ?
<SchrodingersScat>
aboodman: getting pretty zen in here
<whyrusleeping>
jokke: no, you cant. anyone could add any content
<jokke>
Also, if i would accidentally add, say, my private ssh key to ipfs it is automatically compromised or only if i also share the hash with someone? I'm guessing the former since the has will be added to the DHT
<jokke>
*the hash
<whyrusleeping>
jokke: the hash gets shared to the dht when you call add
<whyrusleeping>
but nobody else has the key until they request it
<whyrusleeping>
which potentially could happen very quickly
<SchrodingersScat>
especially if they're checking things sent over the DHT :/
<whyrusleeping>
so in general, its safer to assume its compromised
<jokke>
Mh. That's a serious security concern. A secret key is bad but it could be way worse. Data which, when it saw the light of day, would potentially destroy lifes.
<jokke>
And there's no way of getting it out of ipfs
<whyrusleeping>
jokke: yeah, in general, if you accidentally add data to ipfs
<xelra>
Oh, I didn't know it worked like that. So it's broadcast? That's nice. I don't need to share it specifically? In theory someone could pin the entirety of ipfs then, right? Every time something new hits the DHT?
<whyrusleeping>
kill your daemon quickly
<whyrusleeping>
xelra: yeah, by default, your node announces hashes to the dht
<jokke>
Also a good thing, depending on the use case. Like for whistleblowers or so
<lemmi>
this isn't really a problem specific to ipfs. you put something out there, it can get copied
<Kythyria[m]>
And if it didn't announce hashes, well... you'd still have the "oops I ran `ipfs announce` on something secret" problem, plus you'd keep forgetting to do it when you _do_ want to publish something.
<jokke>
lemmi: you're absolutely right
<jokke>
Yeah
<Kythyria[m]>
lemmi: Indeed. Any time there's a "publish" button it'll get pressed accidentally at some point.
<lemmi>
ipfs' tooling is just better so such cases :P
<jokke>
add -r can be especially nasty
<whyrusleeping>
fortunately, ipfs ignores hidden files and directories by default
<jokke>
Oh ok
<whyrusleeping>
so doing `ipfs add -r $HOME` won't add your .ssh folder (though, there are likely other things in your home directory you don't want shared)
<jokke>
Yeah
<xelra>
Just for understanding better how the system works: So the DHT holds every hash of the whole network. And everyone node strives to have a complete copy of the DHT? Isn't that a size problem at some point? Doesn't the DHT grow steadily? How do things that are no longer available on ipfs at all vanish from the DHT?
<jokke>
Every peer just has a tiny portion of the dht
<SchrodingersScat>
xelra: no, but when I add my tax returns to ipfs, my node sends out the hash, "Yo bros, I got some mad hashes over here." and then the peers know.
<jokke>
With overlaps for redundancy of course
<jokke>
SchrodingersScat: xD
<jokke>
xelra: there is no other data attached to the announcement. Only your ip and port.
<xelra>
I see. So the DHT relies on the fact that every hash is only X nodes away, right? Like 10 nodes maybe to reach every node on the world. How many levels down does such a hash request hop from node to node, if the hash cannot be found? That's what is called TTL, right?
<whyrusleeping>
for ipfs its peerID and multiaddr instead of ip and port
<aseriousgogetta>
peer
<aseriousgogetta>
2 peer
<aseriousgogetta>
;)
<jokke>
So when the irs collects SchrodingersScat, they ask their peers "Yo bros, wheres dat hash?" If they don't know, they ask their peers etc until some peer says "Its at SchrodingersScat raspberry pi"
<jokke>
*collects ... tax returns
<SchrodingersScat>
yes
spossiba has quit [Ping timeout: 255 seconds]
<jokke>
And you can imagine it being fairly fast because the request is a broadcast and each peer then broadcasts again and so on
<jokke>
Spans the globe quite quickly
dconroy has joined #ipfs
spossiba has joined #ipfs
<xelra>
I understand. But it's not asking endlessly, right? Because I might request something that doesn't exist. How many hops does such a hash request live? Or how long? If it's time-limited.
<jokke>
I understand why ipfs is on-demand, but why can't ipns be pushed to a dht?
<whyrusleeping>
ipns references are stored in the dht
<jokke>
Oh!
maxlath has joined #ipfs
<xelra>
So in the Kademlia paper they say 5 hops, if I'm reading that right.
Encrypt has joined #ipfs
<xelra>
I was wondering, because I thought that this might be a possible attack on ipfs.
<xelra>
If I request non-existing hashes on a massive scale, I even get an avalanche effect of multiplying requests through the DHT.
<xelra>
I was wondering whether that's a possible DoS attack.
pat36 has quit [Remote host closed the connection]
pat36 has joined #ipfs
aboodman has quit [Quit: aboodman]
aboodman has joined #ipfs
aboodman has quit [Client Quit]
<voker57>
well you'll have 5 requests to 5 different hosts not 5 to 1 host
<voker57>
not much of help
bitbit has quit [Ping timeout: 268 seconds]
Mateon3 has joined #ipfs
Mateon1 has quit [Ping timeout: 260 seconds]
Mateon3 is now known as Mateon1
upperdeck has quit [Read error: Connection reset by peer]
upperdeck has joined #ipfs
<xelra>
I thought it was 5 levels to about one to ten thousand nodes?
<xelra>
Meaning 10,000^5?
m0ns00n has joined #ipfs
Guest36324 has quit [Quit: quit]
Foxcool has quit [Ping timeout: 276 seconds]
galois_dmz has joined #ipfs
johan_ has quit [Quit: This computer has gone to sleep]
johan_ has joined #ipfs
<whyrusleeping>
other nodes don't make requests for you
<whyrusleeping>
you make a request to a node, get more information, then make a request to another node
galois_d_ has quit [Ping timeout: 258 seconds]
caffix[m] has left #ipfs ["User left"]
mildred has joined #ipfs
johan_ has quit [Ping timeout: 248 seconds]
galois_dmz has quit []
galois_dmz has joined #ipfs
aboodman has joined #ipfs
galois_dmz has quit [Read error: Connection reset by peer]
galois_d_ has joined #ipfs
bnwilson has quit [Ping timeout: 276 seconds]
pepsipeacecorp[m has joined #ipfs
<SchrodingersScat>
xelra: if we tweet about IPFS to elon musk enough, we may just make it on the second flight when the first crew is sick of mars without cat photos.