<Kubuxu>
aah, right, it is after midnight here, sorry for that
<Kubuxu>
I knew something wasn't right
wallacoloo_____ is now known as Wallacoloo
cemerick has joined #ipfs
robattila256 has quit [Ping timeout: 268 seconds]
jonnycrunch has joined #ipfs
cemerick has quit [Ping timeout: 246 seconds]
cemerick has joined #ipfs
jonnycrunch has quit [Client Quit]
jonnycru1ch has joined #ipfs
* jonnycru1ch
* jonnycru1ch
jonnycrunch
jonnycru1ch has quit [Client Quit]
zabirauf has joined #ipfs
subtraktion has joined #ipfs
subtraktion has quit [Client Quit]
subtraktion has joined #ipfs
tilgovi has joined #ipfs
zabirauf has quit [Ping timeout: 256 seconds]
subtrakt_ has joined #ipfs
subtrakt_ has quit [Client Quit]
cemerick has quit [Ping timeout: 246 seconds]
subtrakt_ has joined #ipfs
subtrakt_ has quit [Remote host closed the connection]
jonathan1 has joined #ipfs
subtraktion has quit [Ping timeout: 240 seconds]
shizy has joined #ipfs
matoro has quit [Ping timeout: 240 seconds]
cemerick has joined #ipfs
e337 has joined #ipfs
<kpcyrd>
4.7 also includes the changes to the Docker image that starts deprecating a specific way to start the image, please remind people to check their logs and fix their setups :)
chrisdickinson has quit [Ping timeout: 268 seconds]
Nukien has quit [Ping timeout: 268 seconds]
chrisdickinson has joined #ipfs
Nukien has joined #ipfs
tmg has joined #ipfs
google77 has joined #ipfs
jkilpatr has joined #ipfs
<horrified>
how would you estimate the cost of a full (re-)implementation of ipfs? for example, when someone needs to use ipfs in an environment where it's impossible to get Go. or if someone wants to re-implement ipfs in C as an exercise.
<M-fabrixxm1>
about treefiddy
<M-fabrixxm1>
(sorry)
<horrified>
haha
jkilpatr has quit [Ping timeout: 246 seconds]
<e337>
dumb question but is it feasible for ipfs to be anonymous?
<pinkieval>
horrified: first you would have to implement all the multiformats, which are underspecificied
Guest148421[m] has joined #ipfs
<pinkieval>
that's where I gave up when trying to write my own implementatiopn
<Mateon1>
e337: Currently, no. In the future, maybe. Using Tor is not enough, as IPFS has an ID attached to every node, and even if you reset the peer ID often, people might recognize you by the content you provide
<xloem1>
e337 it's expandible so people can do it over anonymous transports but I don't know that anybody has written the code to. anonymity is hard.
<pinkieval>
Mateon1: that's good if you only want pseudonymity, though
arpu has quit [Ping timeout: 260 seconds]
tkorrison[m] has joined #ipfs
toppler has quit [Remote host closed the connection]
toppler has joined #ipfs
jkilpatr has joined #ipfs
zabirauf has joined #ipfs
arpu has joined #ipfs
<e337>
is it possible to end-to-end encrypt from one node to another?
zabirauf has quit [Ping timeout: 240 seconds]
<e337>
I suppose I'm conflating privacy and anonymity a little.
<e337>
Privacy would cover anonymity, i.e., anonymity is the privacy of origin of the payload, but not privacy of the payload itself.
<r0kk3rz>
e337: you want transport layer encryption?
spilotro has quit [Ping timeout: 256 seconds]
<r0kk3rz>
or something more like TOR?
erikj has quit [Quit: kfonx]
erikj has joined #ipfs
erikj has quit [Changing host]
erikj has joined #ipfs
spilotro has joined #ipfs
<e337>
I was just wondering if Node A can encrypt something so only Node B can decrypt it, but I suppose that can easily be done outside of ipfs.
<r0kk3rz>
e337: sounds like cjdns to me
<e337>
Hmm, cjdns.. Interdasting
Captain_Beezay has quit [Excess Flood]
<e337>
It says here in the ipfs whitepaper that object-level cryptography is supported.
<pinkieval>
What are the possibilities for whitelisting content served by the gateway?
cemerick has quit [Ping timeout: 246 seconds]
cemerick has joined #ipfs
galois_d_ has quit [Remote host closed the connection]
cemerick has quit [Ping timeout: 246 seconds]
gruu has joined #ipfs
ph12[m] has joined #ipfs
ecloud has quit [Ping timeout: 256 seconds]
Boomerang has quit [Quit: Lost terminal]
Boomerang has joined #ipfs
<lgierth>
kpcyrd: could you PR a note about the docker thing to the changeloG?
<kpcyrd>
lgierth: I'm going to prepare something this evening :)
ecloud has joined #ipfs
<lgierth>
awesome, thanks
<e337>
Can someone please explain if and how the ipfs model is different from git?
<SchrodingersScat>
it's in the cloud
<e337>
I can see one or ways so far 1) dht for object distribution 2) bitswap for fairness/incentive
<e337>
For 2, I don't quite understand why anyone would want to host any of my files, ever.
<SchrodingersScat>
i'm not sure i hold data I don't want
<e337>
You're not sure that you hold any data that you don't want? Or you're not sure, and you do hold data that you don't want.
<SchrodingersScat>
correct
<r0kk3rz>
e337: think of things which are files but we current consider ephemeral, like webpages
<google77>
e337 the logic for 2 is that if they wanted to see your cat meme, it also means they will redistribute it after they saw it.
<e337>
Nothing's ephemeral in ipfs either except the ipns, which is analogous to a ref in git, e.g., the curren tHEAD.
<lgierth>
SchrodingersScat: it'
<lgierth>
SchrodingersScat: it's not in the cloude
<lgierth>
please don't confuse people
<SchrodingersScat>
cloud is a buzzword anyway :(
<lgierth>
yeah, so don't use it
<e337>
But why would two completely separate and possibly competing organizations want to host each other's files?
<lgierth>
unless you do mean the cloud ;)
<SchrodingersScat>
and we agree there is no cloud?
<lgierth>
yes
<SchrodingersScat>
ok, good.
zabirauf has joined #ipfs
<SchrodingersScat>
e337: I'm a bit new to ipfs, but I didn't think I held things that I didn't request at some point. Either through pinning or file requests through IPFS.
<lgierth>
yes ipfs only stores what you add yourself, or tell it to fetch from the network
<google77>
also git usually still needs some kind of server, at leat the way most people use it. Remember what panic happened when github was offline one day?
<google77>
*least
<e337>
Well, if nobody else wants to store my files for me, what good are they?
<lgierth>
filecoin is gonna provide an economy on top of ipfs around storing stuff for other people
<e337>
You can run git fetch from any node to fetch objects from any other node.
<SchrodingersScat>
without that, ipfs still seems like a great file distribution method. If your files are 'unpopular' they sit on your server, otherwise they spread as needed, like a virus.
<lgierth>
ipfs doesn't come with a way of "storing stuff on the network" of its own so that these two concerns are cleanly separate, and you can build anything on top of ipfs
<victorbjelkholm>
lgierth: ah, pincoop is very old, haven't looked at it in a while so please don't be surprised if it doesn't work anymore
<victorbjelkholm>
idea is sound though :)
<e337>
Yeah, ok, filecoin, so how much would it cost to host one of my files forever?
<google77>
e337 I know you can.
<lgierth>
victorbjelkholm: yeah thought so ;) does it still require opening up the api to others? ;)
maxlath has joined #ipfs
<SchrodingersScat>
lgierth: if I put my .ipfs directory inside a fuse mounted amazon cloud storage directory is it in the '''cloud'''?
<r0kk3rz>
e337: forever is a long time, so expensive
<victorbjelkholm>
lgierth: it starts it's own cluster with docker containers, so is contained that way, doesn't touch your existing setup. I started writing a golang port that starts it's own daemon and stuff but didn't come very far with that
<e337>
Schrode, only the most popular stuff survives?
<e337>
Future historians will remember us only by Kardashian episodes
zabirauf has quit [Ping timeout: 260 seconds]
<r0kk3rz>
lol. obviously someone has to host it somewhere
<SchrodingersScat>
e337: that's my understanding. or at least how I would use it.
<victorbjelkholm>
e337: there is many different ways you can make stuff survive. For example, if you're building a P2P Twitter, you could make it so you reshare all the content coming from the people you follow
<r0kk3rz>
if nobody hosts it, then it gets forgotten
<victorbjelkholm>
or you have groups, that all agree to share each others content
<e337>
It doesn't really feel that "permanent" if you have to be popular to survive.
<lgierth>
no, not "only the most popular stuff survives"
* lgierth
walks away
<SchrodingersScat>
lgierth: meh
<victorbjelkholm>
lgierth: stop being so clear
<google77>
e337 it's a lot more permanent than http
<SchrodingersScat>
lgierth: if other people don't pin it then seems like it's at risk
<e337>
Especially considering that some categories of archived material tends to be ignored for decades before being "rediscovered"
<r0kk3rz>
e337: its permanent as in file X will always have address Y, not permanent as in forever available
<e337>
google, yeah, but I think that comes from the idea of immutability
<lgierth>
yeah if nobody's hosting it, nobody's hosting it. that's VERY different from "only the most popular stuff survives"
<SchrodingersScat>
k
<e337>
git handles immutability through content-addressable data, and it also handles durability by keeping the entire history in every clone.
<SchrodingersScat>
so long as one person is hosting it, it has unlimited growth potential, like working at a Panda Express
<r0kk3rz>
so like, if the internet archive wants to archive some stuff and the original host dies, nobodys links break
risk has quit [Quit: Connection closed for inactivity]
<e337>
I think generally nobody is going to want to host other people's data at all
<SchrodingersScat>
maybe if your git sucks
<e337>
Just a thought, maybe filecoin or Ethereum can address that.
<r0kk3rz>
yep. because theres certainly no examples of anyone hosting anything for anyone else ever
<r0kk3rz>
:P
<google77>
e337 why not?
<e337>
Well
<e337>
If people can see garbage that's not theirs in their repos, they'll just delete it.
<lgierth>
<e337> I think generally nobody is going to want to host other people's data at all <--- i think history is proving otherwise, has been, and will be
<e337>
i.e., an infinitely greedy node
<lgierth>
if i'm using a certain dataset, that means i have an interest in it, and ipfs makes it available to others too
<r0kk3rz>
e337: how does TOR survive then exactly?
<google77>
also it isn't about wanting to host. It's simple, if you want to see it you gotta host it until it gets out of your cache. An exception is if you pin it and want to keep it indefinitely.
<r0kk3rz>
why would anyone want to route anyone else packets?
<SchrodingersScat>
greedy nodes want the most data
<e337>
tor survives because none of the nodes know what they're transporting or who it belongs to
<e337>
only the exit node has a clue about the content, and only the entry node has a clue about the origin
<google77>
e337 ipfs will not fill the repo with garbage that is not theirs. ever. it will only fill repo with garbage they specifically wanted.
<e337>
Well that won't work if you're aiming for a "permanent" web.
<google77>
ipfs is not freenet, what you are describing is freenet.
<e337>
Obviously people will just delete old shit because it's not "wanted" anymore.
<SchrodingersScat>
if you're the only one that cares about it, then you have to host it
<e337>
That sounds like the way it already is..
<SchrodingersScat>
google77: pretty much, or a tahoe scenario with open access
<e337>
What am I describing?
<SchrodingersScat>
insanity
<e337>
I think a fundamental problem is that objects can, in the general "unpinned" condition, drop out of existence in an unpredictable way.
<r0kk3rz>
why is that a 'fundamental problem'
<e337>
Well because it isn't immutable or permanent.
<e337>
Also the very notion of garbage collection seems odd in a "permanent" web.
<SchrodingersScat>
the link is, like they mention. the link should always point to the same data, whether its there or not
<e337>
Maybe I'm misthinking it
<SchrodingersScat>
I love garbage collection, I ran one today
<r0kk3rz>
sure it is, hash x points to file y, and will *always* point to file y
<r0kk3rz>
its immutable
dimitarvp has joined #ipfs
<e337>
Maybe, if you successfully "pin" something, but why would anyone else want to "pin" your stuff?
<r0kk3rz>
sure the contents of the network can and does change, but the links are deterministic
<SchrodingersScat>
e337: why are you hosting it on ipfs if no one wants it?
<keks[m]>
e337: because e.g. they like my content and want to be able to access it oflfine
<M-fabrixxm1>
*what *that
<r0kk3rz>
e337: IPFS is not storage, its a transport layer :P
<google77>
e337 perhaps he finds it interesting and wants to keep it for later?
<e337>
Well, "no-one wants it" is obviously the worst case scenario which I have to consider, but it could easily happen for very rarely access objects.
<r0kk3rz>
like bittorrent
<lgierth>
the internet archive usually wants everything
<SchrodingersScat>
they take it all
<lgierth>
so there you have at least one institution which will soak up everything that others don't bother to
<r0kk3rz>
or you get public institutions like libraries helping to host wikipedia .etc
<lgierth>
you gotta distinguish between individual users running ipfs on their computer, and large institutions like archives and libraries, who dump a ton of resources into ipfs
<e337>
Ok, well, if you depend on internet archive to be the only one that has *everything*, then that's just centralizing permanence.
<lgierth>
ok i got more important things to do, have said everything i have to say
<google77>
Or you set up an arrangement with your friends and neighbours to pin each others stuff
<lgierth>
it's not a lot of fun discussing with you
<M-fabrixxm1>
e337: you can't have permanent free unlimited hosting from thin air
<r0kk3rz>
e337: you need to stop thinking that permanence == availability
<e337>
Sorry these questions just trouble me
<keks[m]>
let's not forget another reason is that i pay them to do so
<SchrodingersScat>
e337: you can make a fairly robust system with limited resources though. Even using a home connection you could ensure that the files at least exist on the network. Speed and performance would be questionable but like bittorrent it should scale well.
<lgierth>
ok -- well you HAVE to rely on someone. be it a big institution, or people getting together to pool their resourceas
<lgierth>
individualism doesn't help if you wanna build a big network of data
<lgierth>
you gotta rely on harddisk vendors continuing to sell harddisks too
<e337>
Why bother with ipfs if YOU are still completely responsible for ensuring the permanence of your data?
<lgierth>
i think you'll enjoy freenet more if you just want to dump your data into the network
<SchrodingersScat>
^
<google77>
e337 because you are only in the worst case scenario.
<e337>
You could answer, well, the torrent network potentially spreads the bandwidth around.
<M-fabrixxm1>
until you find people that are ok with using part of their hd to store your data in freenet..
<e337>
i.e., you get the advantages of the bittorrent for distributing very popular files, like movies with Will Smith
<SchrodingersScat>
ew
<e337>
jk
<r0kk3rz>
e337: because its a nice toolset for creating distributed applications without having to roll your own p2p layers
<clownpriest>
isn't filecoin supposed to solve all these problems?
<SchrodingersScat>
if they'll ever quit pre-mining long enough to roll it out
<e337>
Yeah, ipfs is definitely cool
<lgierth>
clownpriest: yes i mentioned filecoin but people just keep on mumbling "but but but"
<google77>
e337 ipfs offers you many other benefits, like latency removal, you get distributed web and distributed applications, immutable links etc... IPFS is NOT ONLY ABOUT PERMANENCE
* lgierth
gonna go get some coffee
<clownpriest>
lgierth: sorry, didn't notice your mention under all the mumbling
<lgierth>
yeah ;)
<r0kk3rz>
you dont even need filecoin to pay someone to host your stuff
<M-fabrixxm1>
but I don't want to pay :-P
<r0kk3rz>
pay a vps provider, stick on ipfs, done someone else is hosting your stuff
<e337>
How is ipfs about permanence at all?
galois_dmz has joined #ipfs
<e337>
To me, permanence is hugely important.
<M-fabrixxm1>
ok.. please stop writting "permanent".. it trigger the bot..
<SchrodingersScat>
e337: like right now I'm using ipfs for mostly a 1:1 transfer, but if anyone in my group needs the same stuff then it'll get to them quickly, especially if they call it through the ipfs.io gateways, then hypothetically it's temporarily on those machines as well.
<r0kk3rz>
e337: do you not understand how the links are permanent?
<e337>
To me, permanence is SERIOUS, i.e., what happens to your data if all major US cities were simultaneously nuked?
monkhood_ has joined #ipfs
<SchrodingersScat>
not the same thing afaict
<keks[m]>
it's permanent in the sense that the identifiers are permanent, i.e. independent from the location. that is in stark contrast to the old web
<M-fabrixxm1>
then your data loss is the last of your problems
<SchrodingersScat>
e337: so rent a machine in canada, rent a machine in russia, rent a machine in hong kong, pin all the same stuff to them
<keks[m]>
of course someone still needs to store that data
<keks[m]>
but it doesn't matter who
monkhood has quit [Ping timeout: 240 seconds]
<clownpriest>
nothing is "permanent", as long as entropy is at work
<keks[m]>
links won't break anymore if the content move
<keks[m]>
only when it is gone
<SchrodingersScat>
clownpriest: is entropy permanent?
<clownpriest>
yerp
<e337>
If I have to set up all those machines, what benefit is ipfs (for permanence)?
<r0kk3rz>
you dont
<r0kk3rz>
but someone does, obviously
<M-fabrixxm1>
e337: transport
<SchrodingersScat>
the benefit would be surviving a nuclear holocaust with minimal effort
<google77>
e337 it's about permanence in a sense that its a lot more permanent than http. ins uch a situation I doubt that even freenet would survive.
<clownpriest>
SchrodingersScat: entropy lives in the platonic realm tho, not in the world as is
galois_dmz has quit [Remote host closed the connection]
<e337>
Ok, so if you take away permanance, you're left with a (very nice) dht + bitswap + bittorrent?
tmg has quit [Ping timeout: 260 seconds]
<r0kk3rz>
e337: when you say 'ipfs get file X' you dont care where it comes from, could come from 10 places at once
<r0kk3rz>
if anyone on the network has that file, you'll get it
<google77>
+ a technology that is unlike bittorrent built to be upgraded by further tech, built to servewebsites as well as general files and web services
<e337>
Well, what if nobody has it?
<M-fabrixxm1>
I really whant to know from where you think permanence as "whatever will happend 'til asteroids on earth, my data will be there" came from?
<keks[m]>
identifiers are permanent, data is not
<r0kk3rz>
e337: you know the answer to this question
matoro has quit [Ping timeout: 240 seconds]
<r0kk3rz>
if nobody has the bits, they cannot get sent, obviously
<M-fabrixxm1>
bakups, e337 . backups.
<keks[m]>
well in that case a magic fairy will grant you a golden usb flash drive
<e337>
So since I've been criticizing ipfs a little on the permanence thing, I can offer an idea. Maybe it's terrible..
<SchrodingersScat>
e337: one cool thing about this seems to be that if you want to check if a file is on ipfs you can just hash the file and then check to see if any nodes have it without divulging the actual data.
<e337>
I think a fundamental key to fair sharing is that nobody shoudl actually know *what* they're sharing or *whom* it belongs to.
pippo has joined #ipfs
<r0kk3rz>
e337: you seem to be asking for guarantees that are literally impossible
<keks[m]>
that is not impossible at all
<e337>
Schrode, could that also be a security hole?
<google77>
Maybe one cool thing for you, it doesn't solve storage permanence, but you might still like it. If a file has disappeared from IPFS and someone reuploads it, it will have the same hash.
<keks[m]>
but its kinda pointless
<SchrodingersScat>
e337: seems ok to me, I scanned ipfs for all my nudes, not a one on there.
<M-fabrixxm1>
e337: why?
<e337>
Well, the only downside to total object anonymity is that a malicious actor can dump an infinite amount of garbage into the network.
<keks[m]>
well, depends on the use-case i guess - e337 why is that a fundamental key?
<keks[m]>
you mean spam?
<keks[m]>
i don't think that will be a problem /s
pippo has left #ipfs [#ipfs]
<e337>
google, no, it would, because many duplicates of every chunk of every file would be scattered throughout all nodes equally, with increasing probability to favor optimal paths for distribution.
<google77>
e337 freenet already does what you suggest. The issue is that commercial companies will never tolerate that, as they would never want to have potentially illegal stuff on there, even if it's encrypted. So it's a legal problem.
<e337>
Well, that seems very iffy.
<keks[m]>
doesn't freenot only store stuff of you and your friends?
<e337>
I think we are going to have to get past the whole encrypted/illegal thing.
<SchrodingersScat>
keks[m]: thought it had a mode where it could access a public grid
<e337>
I think the network needs to be 100% anonymous in order for it to be really permanent.
<google77>
keks i think companies would only be interested in the public part of freenet...well mostly.
<r0kk3rz>
even then its not really permanent...
<r0kk3rz>
you'll still lose bits
<keks[m]>
i guess
maxlath has quit [Ping timeout: 240 seconds]
<google77>
anonimity is a whole other can of worms.
<r0kk3rz>
e337: anyway, what you describe is not what IPFS is or wants to be :D
<e337>
Perfect anonymity is important to guarantee that all data is spread equally everywhere, thereby ensuring the survival of *all* data.
<keks[m]>
e337: remember how the gentoo wiki died? the datacenter company was bankrupt, and all the backups were in the same datacenter. the same would have happened if it would have been hosted on an onion address
<SchrodingersScat>
i like this guy, he's funny
<keks[m]>
e337: what does anonymity have to do with spreading the data?
caiogondim has joined #ipfs
<google77>
e337 you know that ipfs as it is now is already a extremely ambitious project and your ideas are aare at least 10 times more ambitious than that?
<r0kk3rz>
keks[m]: or more recently with cyanogenmod...
<e337>
Well if you put node operators in charge of picking and choosing what to host, then there will inevitably be huge holes in how certain files are spread.
<r0kk3rz>
of course
<e337>
Taken to its extreme, node operators won't host anything they don't specifically want, which in the case of a business, will usually be *anything* they didn't produce themselves.
<e337>
So you might end up with many separate balkanized networks..
<keks[m]>
the networks won't be balkanized. you will still be access the stuff from the other end of the network
<M-fabrixxm1>
taken to its extreme, no-one will host anything.
<e337>
i.e., a private ipfs for each company, but it would be private between the client and the company, i.e., it degrades to exactly what we already have
<r0kk3rz>
the 'network' is not guaranteed to even be a single cohesive thing
<r0kk3rz>
you could have many ipfs networks
<google77>
I mean you suggested PERFECT ANONIMITY, which is an issue that many inteligent people have been working on for 20-30 years, and they still couldn't figure it out. And you added storage permanency to that.
<r0kk3rz>
its distributed after all
shizy has joined #ipfs
<keks[m]>
r0kk3rz: that isn't very likely though
<e337>
Well, storage permanency I think is a direct result of perfect anonymity, because node operators can't tell anything about a particular block, they can't be biased in terms of what they choose to host.
<e337>
Therefore every block is distributed equally.
<keks[m]>
anyone in the networks could start cross-announcing and the networks would merge
<r0kk3rz>
keks[m]: yes it will heal itself automatically once the physical network connectivity resumes
<r0kk3rz>
but cmon, you've never been somewhere and the internet has gone down?
<e337>
You'd just have a certain amount of redundancy for each block and continually balance for high or low demand situations.
<r0kk3rz>
you've never heard of a country cutting off outside internet access?
<keks[m]>
r0kk3rz: i life offline first
<keks[m]>
*ive
<M-fabrixxm1>
e337: and when your new project will be avaiable? :-P
<SchrodingersScat>
r0kk3rz: egypt :(
<keks[m]>
e337: a content creator will always be able to identify their own content
<r0kk3rz>
exactly
<e337>
I'm just brainstorming, maybe I'm just being batshit
<r0kk3rz>
e337: you could implement what you're describing as some kind of service on top of ipfs
<e337>
Well, if the owner can tell it's his, then he can simply choose not to host anything that's not his.
<google77>
e337 if it's brainstorming then it's fine. I however do think your ideas are highly unrealistic at this point in the development of computer science. At least what my research shows.
<google77>
Maybe in 20-30 more years.
<r0kk3rz>
e337: your magical network doesnt exist in a vacuum, people have to choose to join
<e337>
Ok
<keks[m]>
e337: i know that this is your point, but i'm saying you won't be able to stop a content creator from from identifying their own content
<keks[m]>
at least not with 2017 production-level crypto
<e337>
So each node would have peers that act as block storage responsibility police. Like a neighborhood watch that continually challenges the node to prove he still has blocks that he should have.
<google77>
e337, don't please be discouraged of thinking, your ideas were creative and thinking is a good thing to do.
<e337>
Bingo! Solved!
<r0kk3rz>
keks[m]: would information encoded in such a fashion even be retrievable?
<keks[m]>
depends on your crypto I guess?
<e337>
The nice thing about the "neighborhood watch" idea is that it also does constant verification for data integrity, thereby ensuring *permanence*.
<r0kk3rz>
i dont think crypto comes into it, this is information theory
<keks[m]>
e337: yeah I guess that would work, but proving ownership to someone who doesn't own a file is tricky as well
<keks[m]>
r0kk3rz: and so is crypto
<e337>
The question is whether you can deterministically tell which blocks any given node *must* host.
<e337>
Then challenge that node into producing the blocks they are responsible for hosting.
<e337>
Although what would you do if they fail the challenge?
<e337>
(Just to clarify, a node knows *some* of his blocks are his, but nobody else knows that.)
<keks[m]>
yup, and what do you do if they run out of storage?
<e337>
Well, they have to advertise a certain allotment, and their allotment gets filled up with everyone else's blocks.
<e337>
Everyone knows how much storage the node has, so they know it can fit X blocks, so there better be X blocks in there.
<e337>
I know what we can do to punish a rogue node..
<e337>
EVERYBODY DDOS!
<e337>
I still don't know how to prevent a rogue node from filling up the network with random garbage..
<keks[m]>
e337: collective resource allocation is not a new problem. there have been a few proposed solutions
<e337>
Actually, the more garbage and noise is pumped into the network, the more anonymous it becomes.
<keks[m]>
a) capitalism: let the market handle it. I think this is filcoins approach
<lgierth>
that ^ is very very simplified
<lgierth>
but yes, market
<keks[m]>
b) dictatorship: you have a central node dictating who stores what
<e337>
Technically, in order to obfuscate the origin of a block, you'd need a constant flow of noise and garbage throughout the network.
<keks[m]>
c) solidarity: this requires a certain amount of trus between the nodes (that they don't lie to each other)
<e337>
i.e., each node consuming 100% bandwidth 24/7 just sending random noise
<keks[m]>
if you have trust between the nodes, you might not need "block anonymity"
<e337>
google, thanks, I will definitely check out freenet though.
<SchrodingersScat>
e337: you may also be interested in Tahoe-LAFS
<keks[m]>
lgierth: yeah I figure. i haven't seen much docs about it yet
<keks[m]>
much docs, such wow
<e337>
Yeah, a) and c) seem to be at odds
<e337>
I don't think b) would work if the dictator were simply overthrown.
<e337>
I'd say a combination of a) and c) is how it's already done now. a) for paying for your own servers and storage, and c) for a shared, semi-communal internet backbone system
<lgierth>
keks[m]: just saying simplified because capitalism is much more than just "hurr durr market"
<e337>
Assuming things like "internet backbones" are still important
<keks[m]>
lgierth: granted
<lgierth>
:)
<e337>
Yeah, maybe you just pay random people to host and distribute yoOCur encrypted blocks for you.
<e337>
Like airbnb for your hard disk and internet connection..
<e337>
*looks into filecoin*
<e337>
Anyway, it's "permanent" until you run out of coin.
<clownpriest>
e337: the market ends up deciding if its worth continuing to host your data. if your data is valuable, there's incentive for strangers to host it
<lgierth>
the market doesn't decide anything :) the market is just a metaphor
zabirauf has joined #ipfs
zabirauf_ has joined #ipfs
<e337>
Byteball is also an interesting decentralized storage system
<clownpriest>
lgierth: sure, but the object of the metaphor "acts" as if there is volition, in so far as it has any effects
<Mateon1>
Well, with Filecoin, due to how free market works, people will find the most effective ways to store the most data in the cheapest way possible. I bet you, in 5-10 years, you'll be able to lease a terabyte of storage for a month for a few dollars.
<Mateon1>
Also, there are people who will host other people's stuff, for free even
<e337>
Is there like a Filecoin to BTC exchange?
<Mateon1>
Look at archive.org
<clownpriest>
just like water "decides" to boil at 100 celsius
<lgierth>
e337: filecoin doesn't exist yet
<lgierth>
clownpriest: fair enough, but markets are not nearly as scientific as "water boils at 100 celsius" ;)
<r0kk3rz>
e337: say permanent again, i dare you, no i double dare you :P
<lgierth>
much of economic science isn't worth the name science
<pinkieval>
lgierth: actually, water does not always boil at 100 celsius
<lgierth>
yeah, depends on pressure ;)
<e337>
8(o)*WRONG*
<clownpriest>
lgierth: personally i consider economics a pseudoscience. but human behavior, of which the "market" is comprised is very much characterizable
<lgierth>
agreed on that
<e337>
Everything must be designed to survive a multiple nuclear blasts nowadays.
zabirauf has quit [Ping timeout: 240 seconds]
<e337>
That's what made the internet so resilient--it was designed with the threat of nuclear war in mind.
<r0kk3rz>
why? seriously. if nuclear war happens we have *much* bigger issues than ones and zeros on computers
<e337>
So the *network* may survive, but if you have too many eggs in too few baskets (Recent Amazon EC2 Outages!), you will suffer great losses.
<r0kk3rz>
ipfs helps with that, you might not think so but it does
caiogondim has quit [Quit: caiogondim]
<e337>
Yes, you may have bigger problems, e.g., possibly being dead, but you don't want to have to deal with nuclear war *AND* your favorite website being down at the same time.
<pinkieval>
e337: I'm thinking of creating a free pinning service for IPFS (available only via cjdns)
<r0kk3rz>
the horror
<pinkieval>
not sure if that will work, though
<pinkieval>
the idea is to have multiple people offer drive space, so it's well distributed
<e337>
What better way to pass the time in your comfy vault than skyping your buddies
google77 has quit [Quit: leaving]
<e337>
pinkieval, interdasting
ZaZ has joined #ipfs
<e337>
A high-trust network must be exclusive.
<pjz>
I've got a for-pay pinning service
structuralist has joined #ipfs
<e337>
Wouldn't you want to auto-pin everything you add yourself?
<e337>
Or do you not care about historical revisions?
<e337>
You'd lose the ability to go back in time to any point unless everything were always kept pinned.
<e337>
Incidentally, is there a "history" mechanism in ipfs?
<e337>
Cause, you know, not to bring up git again, but..
<e337>
Also there's this little thing called branching and merging.
<pinkieval>
e337: what do you mean?
<pinkieval>
isn't everything auto-pinned already?
<e337>
If you don't pin a block, it could be dropped later even though you already added it.
<e337>
I thought you have to explicitly pin things if you don't want them garbage collected..
<e337>
Anyway the idea of pinning seems odd to me, but maybe I'm just not understanding the point of it.
<r0kk3rz>
it autopins everything you add to a node
<e337>
Ah, thanks.
<pinkieval>
e337: when you browse the internet, you don't necessarily want to pin everything you see
<pinkieval>
eg. ipfs cat blah > debian-dvd.iso
<r0kk3rz>
not pinning gives you a running cache, you can say allocate 10gb to that cache
<r0kk3rz>
but if you reach the limits, it will garbage collect
<e337>
Ok
zabirauf_ has quit [Remote host closed the connection]
<e337>
It sounds like there would want to be master nodes that always pin "the whole internet"
<e337>
Then a bunch of caching nodes
<e337>
Or something
zabirauf has joined #ipfs
<r0kk3rz>
i see no reason for that
<r0kk3rz>
no one system runs the 'whole internet' now
<r0kk3rz>
but ideally, yes there will be 'servers' who pin content and 'consumers' who cache
<pinkieval>
pjz: out of curiosity, what volume of data is pinned by pinbits?
zabirauf has quit [Ping timeout: 256 seconds]
chris613 has joined #ipfs
cemerick has joined #ipfs
matoro has joined #ipfs
cemerick has quit [Ping timeout: 246 seconds]
cemerick has joined #ipfs
caiogondim has joined #ipfs
espadrine_ has joined #ipfs
jkilpatr has quit [Quit: Leaving]
espadrine has quit [Ping timeout: 264 seconds]
Jesin has quit [Quit: Leaving]
ylp has quit [Quit: Leaving.]
Jesin has joined #ipfs
ulrichard has quit [Remote host closed the connection]
espadrine_ has quit [Ping timeout: 256 seconds]
onabreak has quit [Ping timeout: 260 seconds]
jkilpatr has joined #ipfs
guest147 has joined #ipfs
onabreak has joined #ipfs
<guest147>
Why can I access the webui from other pc in my network?
<guest147>
cant*
<whyrusleeping>
guest147: because its restricted to only being able to be accessed by your computer
<guest147>
can i enable it so i can access it from my local network?
<whyrusleeping>
You can, yes. It should work if you edit your .ipfs/config Addresses.API field to listen on your LAN ip instead of 127.0.0.1
<whyrusleeping>
and something about CORS, daviddias[m] how do CORS work again?
<whyrusleeping>
guest147: check the console for errors
<guest147>
18:33:57.797 ERROR core/serve: Path Resolve error: no link named "webui-he.json" under Qmc6bAsXTFFdvV3RoLuf6jjV3jzsRLntfsivWQZ75s8QaL gateway_handler.go:534
<whyrusleeping>
what address are you viewing it through? localhost:5001/webui? or from another computer?
<guest147>
another pc
<whyrusleeping>
does the browser console print anything weird?
<guest147>
yes some weird errors
<whyrusleeping>
such as?
<guest147>
SyntaxError: Unexpected number in JSON at position 4 at JSON.parse (<anonymous>) at ipfs-webui.0.3.0.js:14 at n (ipfs-webui.0.3.0.js:2) at r.<anonymous> (ipfs-webui.0.3.0.js:17) at r.n.emit (ipfs-webui.0.3.0.js:2) at E (ipfs-webui.0.3.0.js:55) at y (ipfs-webui.0.3.0.js:55) at ipfs-webui.0.3.0.js:2 at u.run (ipfs-webui.0.3.0.js:1) at s (ipfs-webui.0.3.0.js:1)
<victorbjelkholm>
guest147: seems like it's trying to load a translation that we don't have in webui yet
<victorbjelkholm>
it's trying to resolve :hash/webui-he.json which results in a 404, returning a message that "merkledag couldnt be found" or something like that and webui is trying to parse that into a js object, expecting json
<guest147>
whyrusleeping: just copy the commands there?
Guest71165 has joined #ipfs
<guest147>
whyrusleeping: now I have other error "ipfs-webui.0.3.0.js:12 DOMException: Only secure origins are allowed (see: https://goo.gl/Y0ZkNV).";
tilgovi has quit [Ping timeout: 246 seconds]
ygrek_ has joined #ipfs
Zbyszek[m] has joined #ipfs
<whyrusleeping>
guest147: this is beyond what i know about browsers and webapps
<whyrusleeping>
victorbjelkholm: any idea?
ZaZ has quit [Quit: Leaving]
tilgovi has joined #ipfs
weiwei has joined #ipfs
<weiwei>
hi guys, there is a way to instantiate multiple IPFS nodes with the same ID?
<Kubuxu>
weiwei: I don't think it is good idea, why would you like to do that?
<Kubuxu>
guest147: what link are you using to visit webui?
matoro has quit [Remote host closed the connection]
ashark has joined #ipfs
jkilpatr has quit [Quit: Leaving]
jkilpatr has joined #ipfs
matoro has joined #ipfs
monkhood_ is now known as monkhood
kulelu88 has joined #ipfs
stoopkid has joined #ipfs
twist_ has joined #ipfs
<twist_>
Hello. I am trying to set up IPFS. Could anyone tell me why does http://localhost:5001/webui does not show up?
<twist_>
ERROR core/serve: Path Resolve error: Failed to get block for QmQCSYBkVKY2Am8WU8iJ2i2qYB59TyiPHs9ZxASuxsEkbH: context canceled gateway_handler.go:534
<twist_>
possibly point me to a source? or tell me where I am wrong?
<lgierth>
it's trying to fetch the respective data (html/js/css/images)
<lgierth>
context cancelled just means you cancelled the http request
<lgierth>
one of those errors that aren't actual errors :)
<twist_>
started the server again :
<lgierth>
data retrieval is bound to the http request context, so the daemon doesn't accumulate stale longrunning queries over time
<twist_>
Umm... Okay. I restarted daemon via `ipfs daemon` . Went to browser and pasted http://localhost:5001/webui ... got error = 05:47:37.140 ERROR core/serve: Path Resolve error: Failed to get block for QmPhnvn747LqwPYMJmQVorMaGbMSgA7mRRoyyZYz3DoZRQ: context canceled gateway_handler.go:534
<lgierth>
just what i just told you
<lgierth>
it's fetching the data
<lgierth>
you abort the request, it'll stop fetching the data ("context canceled")
<lgierth>
the code for the webui isn't part of go-ipfs, so it has to fetch it the first time you use it
<twist_>
I din't abort the request. (from terminal or the browser.) What should I do? ( I am missing a lot of context about go-ipfs here. I have just started ipfs three hours ago )
s_kunk has joined #ipfs
<lgierth>
mh, weird. how long does it roughly take until that error appears? timeouts for that are 1 hour, "context canceled" usually hints at the request just getting aborted
<lgierth>
any errors in the browser console?
<lgierth>
might have to try firefox instead of chrome
<lgierth>
gotta run o/
matoro has quit [Ping timeout: 240 seconds]
<twist_>
I am behind a proxy. That could be a possible source of error? (I am bypassing proxy for local addresses, though)
skeuomorf has joined #ipfs
<twist_>
Firefox does not work ... The error comes within a minute.
<twist_>
Thanks @lgierth
<twist_>
Anyone else facing same problem?
Encrypt has quit [Quit: Quit]
twist_ has quit [Ping timeout: 260 seconds]
Boomerang has quit [Remote host closed the connection]
Aranjedeath has joined #ipfs
kulelu88 has quit [Ping timeout: 240 seconds]
<pjz>
pinkieval: not enough ;)
shakalaka has quit [Quit: bye.]
ShalokShalom has joined #ipfs
shakalaka has joined #ipfs
mbags has quit [Remote host closed the connection]
Jesin has quit [Quit: Leaving]
jokoon has joined #ipfs
subtraktion has joined #ipfs
subtraktion has quit [Remote host closed the connection]
zabirauf has joined #ipfs
subtraktion has joined #ipfs
kulelu88 has joined #ipfs
robattila256 has quit [Quit: WeeChat 1.7]
Jesin has joined #ipfs
zabirauf has quit [Ping timeout: 260 seconds]
realisation has joined #ipfs
robattila256 has joined #ipfs
robattila256 has quit [Client Quit]
chungy has joined #ipfs
jokoon has quit [Quit: Leaving]
jsgrant- has joined #ipfs
gde33 has joined #ipfs
apiarian has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
Encrypt has joined #ipfs
Jesin has quit [Quit: Leaving]
matoro has joined #ipfs
monkhood_ has joined #ipfs
monkhood has quit [Ping timeout: 240 seconds]
cemerick has quit [Ping timeout: 246 seconds]
c0dehero has quit [Ping timeout: 255 seconds]
Wallacoloo has joined #ipfs
c0dehero has joined #ipfs
monkhood_ is now known as monkhood
jkilpatr has quit [Ping timeout: 256 seconds]
Jesin has joined #ipfs
kulelu88 has quit [Ping timeout: 260 seconds]
<dryajov>
anyone around and willing to digg through some weird issue with me in libp2p-circuit? :)