<voker57>
good luck getting it out of the net now :)
}ls{ has joined #ipfs
<lgierth>
is it data that's not supposed to be out three?
<lgierth>
i can purge it from the ipfs.io nodes at least
jimjamjim has quit [Ping timeout: 246 seconds]
<clownpriest>
is there a quick way to unpin everything?
Lochnair has quit [Quit: Connection closed for inactivity]
<lgierth>
clownpriest: `ipfs pin ls --type=recursive | ipfs pin rm` should do it
<clownpriest>
lgierth, awesome, thank you
<lgierth>
there's a bug if you have more than 1000-or-so hashes pinned
jimjamjim has joined #ipfs
<lgierth>
i.e. a bug in how `pin rm` then parses the arguments
<lgierth>
having that many pins is totally fine
<clownpriest>
ok cool
gmoro has quit [Remote host closed the connection]
<lgierth>
jimjamjim: is it private data that's not supposed to be online? i can purge it from the ipfs.io nodes at least
<clownpriest>
any chance there are docs on how mdag.Traverse() works? I've been reading the test.go files to try and figure things out, with limited success
<jimjamjim>
lgierth: no.. I was just asking
<jimjamjim>
lgierth: purge it from ipfs.io??
<jimjamjim>
how?
<clownpriest>
i just want to be able to load all the files in the ~/.ipfs repo and traverse dag
<clownpriest>
the dag*
<lgierth>
you'd tell me the hash and i'd do `ipfs block rm` for each block of the hash
<lgierth>
clownpriest: you can use `ipfs refs local` to get all blocks you have, but you'll need to do the graphing yourself
<clownpriest>
there's no interface within the go-ipfs code for doing this? i want to avoid the command line or the http api
<jimjamjim>
lgierth: do you have access to ipfs.io?
<ntninja>
lgierth: Doesn't that mean that any troll could just re-request the data for it to be served from the gateway again?
erictapen has quit [Ping timeout: 240 seconds]
<jimjamjim>
holly cow... I'm talking to Lars Gierth?? the creator of ipfs?!
<ansuz>
:D
<ansuz>
<3 lgierth
<jimjamjim>
ansuz: am I really??
<ansuz>
bah oui
<ansuz>
(that means yes)
Mateon2 has quit [Remote host closed the connection]
Mateon2 has joined #ipfs
<jimjamjim>
:D.. i had no idea.. i just ddg'd 'lgierth' to find the ipfs github repo!!
<ansuz>
heh
<lgierth>
<3 does that mean i'm internet-famous now?
<lgierth>
i'm just one of the developers
<lgierth>
i do have access to ipfs.io, yes
<lgierth>
ntninja: yes -- what is out there once can't be removed with 100% certainty
<lgierth>
we want to eventually have opt-in configurable blocklists, so that individual node operators can decide to not take part in sharing certain stuff
<ansuz>
am I internet-famous by proximity?
<ntninja>
I was just surprised that you don't block content like you do for DMCA'd stuff 😊
<jimjamjim>
lgierth: you are already a legend in my book!! AWESOME WORK!!
<ansuz>
lgierth and I ate poutine together
<voker57>
blocklists will be rushed out when first DMCA hits ipfs.io :)
<Kubuxu>
Does it mean I am also internet famous?
MDude has joined #ipfs
<ansuz>
hi Kubuxu
Encrypt has joined #ipfs
<Kubuxu>
\o ansuz how it is going?
<ansuz>
it's going well
<ansuz>
playing with some crypto stuff
pcctw has joined #ipfs
<ansuz>
and you?
<Kubuxu>
nice, what crypto?
<Kubuxu>
I am working on gateway health tester and new garbage collector for ipfs
<ansuz>
figuring out some general algorithms for p2p decision making
<lgierth>
ntninja: we do, nginx is in front of go-ipfs on ipfs.io, and nginx has rules for a couple of hashes that we block
<ansuz>
I've been thinking about how I'd go about re-building some games without needing an arbitrator
<lgierth>
what i mean is we don't have any solution for this in ipfs itself yet
<ansuz>
sounds like a feature
<Kubuxu>
ethereum is interesting in this regard, you could for example play chess off chain
<lgierth>
jimjamjim: thanks :)
<Kubuxu>
if someone disagrees about the state of the game (you think your opponent is cheating)
<Kubuxu>
you settle it on chain
<ansuz>
yea, there's a chess game you can play on scuttlebutt
<Kubuxu>
otherwise you keep playing off chain with 0 latency
<Kubuxu>
thing is afaik you can't encode rules of chess into scuttlebutt, which means you could cheat
<Kubuxu>
you can encode rules of chess into ethereum
<ansuz>
the current problem I'm figuring out is determining who goes first given a list of players
<ansuz>
there's a bunch of ways to do it, but I want it to be 1) fair 2) deterministic 3) non-interactive
<Kubuxu>
for every player: hash(myname, sort(rest_of_players))
<Kubuxu>
lowest hash wins
<ntninja>
voker57: Why is QmVBEScm197eQiqgUpstf9baFAaEnhQCgzHKiXnkCoED2c blocked on the public gateway then?
<Kubuxu>
or better: hash(myname, sort(all_plpayers))
<voker57>
ntninja: why are you asking me
<ntninja>
Oh lgierth already replied 😃
<voker57>
also you probably shouldn't share warez links on irc
<ansuz>
Kubuxu: that needs a sort that's the same every time
<Kubuxu>
ansuz: it is 1. fair, chance is random, 2) deterministic, all players can check all other players order 3) no communication is needed apart from knowing all the players
<Kubuxu>
> that needs a sort that's the same every time, hmm?
<lgierth>
ntninja: yeah we're not very happy with sharing warez over ipfs
<ntninja>
voker57: Because you answered me first
<lgierth>
and i'd ask you to not paste hashes of it in the future
<lgierth>
clownpriest: yeah it's in ipfs/go-ipfs-api
<lgierth>
lidel: oh awesome \o/ review went through?
ianopolous has quit [Ping timeout: 246 seconds]
ianopolous has joined #ipfs
erictapen has quit [Ping timeout: 240 seconds]
cws has joined #ipfs
<clownpriest>
lgierth, ahh cool. any chance there's an api within the core go repo? some functions i can call on a IpfsNode?
<lidel>
lgierth, yeah but it seems that "manual review" at Chrome Web Store is a joke -- they accepted version that fails to load due to `eval` without CSP with `unsafe-eval` ;-)
<clownpriest>
maybe i just need to let go of the idea of using go-ipfs as a library. but it's such an attractive idea
<lgierth>
clownpriest: i don't think it's properly exposed right now -- search for "floodsub" though, maybe you can get the underlying object somehow. (depending on how, it might break in the future)
<lgierth>
the repo and interface are go-libp2p-floodsub and go-libp2p-pubsub
<clownpriest>
lgierth, cool, thanks
<clownpriest>
are there any people working on search engine protocols on top of ipfs? it's something i've been thinking about for a long time now, wondering what solutions others have come to
<clownpriest>
i saw one project that just connected elasticsearch to ipfs, but that's about it
<lgierth>
Magik6k has built a title search for tr.wikipedia-on-ipfs.org using an ipld data structure as an index
<lgierth>
he can tell you more
<clownpriest>
interesting
<lgierth>
anyhow it always boils down to a static structure and building it in a way that makes lookups easy
<lgierth>
and sometimes you want to be able to update it too, without completely rebuilding it
<clownpriest>
there needs to be a way to version control the index
<lgierth>
and mikolalysenko has really cool ideas for map routing on ipfs
<lgierth>
a la openstreetmap
<lgierth>
well version control is just a log of pointers ;)
<clownpriest>
yeah, just need to be able to bump the pointer on each index update and have a system for reaching consensus on which pointer is the main chain
<clownpriest>
probably doable with crdt's
<lgierth>
if there's only one source of truth, you don't even need the CRDTs
<clownpriest>
though maybe we don't want a main chain
<lgierth>
CRDTs are only neccessary if you want to reliably and provably merge stuff from multiple writers
<clownpriest>
otherwise you might end up with a monolithic system, in which case we might as well just let google index the merkleforest
<clownpriest>
yeh
<lgierth>
large scale cooperative full text search indices on CRDTs
* lgierth
dreams
<clownpriest>
i'mma get it done
<lgierth>
back to cleaning the bathtoom... laters
<clownpriest>
lol peace
dconroy has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<clownpriest>
you can use it to weight words against whole articles
<clownpriest>
maybe you are actually doing this already, need to read more carefully
<Magik6k>
No, this was done pretty much only with the trial-and-error method
<Magik6k>
though it might have some form of tf-idf now that I look at it
<Magik6k>
More popular words have less weight, so pretty much yes
hamersaw has quit [Quit: WeeChat 1.9]
<clownpriest>
popular relative to the whole corpus? or a single article?
<clownpriest>
building forward and inverted indices for all of ipfs shouldn't be too hard. especially since links don't change, don't need to re-crawl like google does
AtnNn_ is now known as AtnNn
chungy has joined #ipfs
dgrisham has quit [Quit: WeeChat 1.7.1]
<Magik6k>
Only the titles are indexed, the more a word appears in title list the less it's popular
<Magik6k>
This approach wouldn't scale well when working with entire articles
<clownpriest>
ahh
erictapen has joined #ipfs
<clownpriest>
wish i could quit my job and work on my engine full time...ugh....if only i could figure out how to not need to eat or pay rent...
<clownpriest>
should probably just buy more bitcoin
dconroy has joined #ipfs
jaboja has quit [Remote host closed the connection]
redfish has quit [Remote host closed the connection]
fred_ has joined #ipfs
fred_ is now known as Guest8301
<gde33>
it would be cool if some firefox hacker created some [low maintanance] feature to prompt the user to install a required software for a protocol. Have some simple messaging between the browser and that software. Preferably something that doesn't involve ugly localhost hax.
<gde33>
The list of potential nice things we cant have is quite long.
<gde33>
one cant realistically require endusers to use a cli to view websites.
jaboja has joined #ipfs
<gde33>
should try get gopher to work, see if you are nerd enough :P
<gde33>
voker57: An uri scheme alone doesn't make something usable for end users?
<voker57>
gde33: yeah, but at least you could give real links to people
<gde33>
no you cant
<voker57>
currently browser extensions us http://*/ipfs/* as convention for redirecting to ipfs daemon
<voker57>
?
<gde33>
what you want is a link like ipfs:// then have the popular browsers prompt the user to install something.
<cyberwolf[m]>
voker57: thx. Tomorrow, I'll check
<gde33>
but if you try make that happen for some singular thing the mozillas wont have any of it. Thats why I think something generic would be nice.
<voker57>
well that's mozilla's hubris
<voker57>
URI syntax is generic enough
<gde33>
its not, not if you want average joe to use something
<voker57>
and browsers already open links like magnet: in apps
dconroy has joined #ipfs
<gde33>
I did install ipfs and read the manual some time ago. I dont recall how to use it.
<gde33>
I now have to read the manual to open a link?
<gde33>
forget it
<voker57>
people had to learn to use http:// links at some point
<voker57>
I don't see a problem
<voker57>
and again, magnet: links exist and work in browsers sometimes
<ntninja>
gde33: There is/was "fs:/ipfs/...", but it got killed in the WebExtension transition
<voker57>
you give average Joe ipfs:// link, if he fails to open it, you direct him to webpage with nice button named "install ipfs"
<gde33>
yeah?
<gde33>
with a picture of a hamster
<gde33>
what is wrong with that?
<voker57>
picture of a hamster?
<gde33>
or some other fuzzy creature
<voker57>
you mean people won't install ipfs to see hamsters?
<gde33>
I think the gopher example is good. There use to be native support in firefox.
<voker57>
thx no firefox is bloated enough already
<voker57>
and a pipe dream anyway
<gde33>
All I want to see is to have all the nice technologies usable for average people so that I can take them seriously.
<gde33>
for ipfs or filecoin there might be enough market share for people to wonder how to use it
<gde33>
but there is so much more
mildred3 has joined #ipfs
<gde33>
indeed the firefox peeps dont want the bloat. So why not have something generic?
mildred1 has quit [Read error: Connection reset by peer]
mildred2 has quit [Read error: Connection reset by peer]
mildred1 has joined #ipfs
<gde33>
I dont see the point of say having every alt coin and all the users jump the hoops to get something as basic as a transaction going
mildred4 has joined #ipfs
<gde33>
I think the whole point of having automation was to not have to do things like that
<DuClare>
The future isn't here yet
<voker57>
There used to be system in KDE called KIO slaves which can be used to view/manage files (and browse websites) through any pluggable protocol
<gde33>
DuClare: but should we build it?
<voker57>
not sure what's going on in kde nowadays maybe it still works
mildred has joined #ipfs
<voker57>
that'd be a viable option for implementing ipfs protocol
<voker57>
it's hella generic
<gde33>
nice
<ntninja>
voker57: That still exists
<DuClare>
gde33: It's probably a social/political problem. I might as well become the president while I solve it :/
<gde33>
"good enough" is the opposite of perfection.
mildred3 has quit [Ping timeout: 248 seconds]
kvda has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
mildred4 has quit [Ping timeout: 240 seconds]
mildred1 has quit [Ping timeout: 240 seconds]
erictapen has joined #ipfs
ShalokShalom has quit [Quit: No Ping reply in 180 seconds.]
larpanet has joined #ipfs
ShalokShalom has joined #ipfs
echoSMILE has joined #ipfs
dconroy has quit [Read error: Connection reset by peer]
erictapen has quit [Ping timeout: 240 seconds]
rendar has quit [Quit: std::lower_bound + std::less_equal *works* with a vector without duplicates!]
<cwahlers_>
i'm with gde33. for a technology to succeed today, i believe that good UX is vital. i think the ipfs team is aware and agrees. ipfs gateways are nice to have for a start, but in the end go against the very idea of a decentralized system
<cwahlers_>
"normal" users don't install daemons on the command line. they use their ipad or their phone, and if there's no app for that or if it doesn't work in their device's browser, that's a no go for mainstream adoption.
<cwahlers_>
what you could do now apart from webextensions is (not a perfect solution) have a service worker that hijacks certain well defined urls (like http urls to a gateway) and serve it via ipfs instead. maybe that would even work on a phone though i have - hopefully unwarranted - concerns about memory and bandwith).
<cwahlers_>
or convince all browser makers to bundle ipfs and add a new url scheme
<gde33>
cwahlers_: maybe all that is needed is a standard for installing helpers and them helpers talking with the browser?
<cwahlers_>
gde33 you mean browser plugins?
<gde33>
if you label it as anything specific it gets a lot less interesting
<gde33>
cwahlers_: something browsers would implement
<gde33>
have a list of formally reviewed software that may prompt the user to install something to be able to use something.
<cwahlers_>
so, browser plugins? :) they just more or less all got rid of that, for.. reasons
<gde33>
something like a plugin platform
<cwahlers_>
or maybe release a whole new "ipfs browser" :)
<gde33>
yeah, lets have command line tools and dedicated browsers for everything that isn't http://
<whyrusleeping>
maintaining your own browser is a lot of work, and generally leads *away* from being adopted in mainstream browsers
<gde33>
frankenstein software
<gde33>
whyrusleeping: exactly, sorry the sarcasm doesn't work over irc
<voker57>
mainstream browsers are frankenstein software
<whyrusleeping>
heh, no worries. I was pretty sure you were being sarcastic