<drathir>
at start im warning im newbe to not angry anyone w/o reason, bc that not my intention... im wonder if somethin changed, bc Getting Started report 4096b keys and fresh install generating 2048b key...
Eudaimonstro has joined #ipfs
<whyrusleeping>
drathir: yeah, we've lowered the default key size and havent updated the docs yet
<whyrusleeping>
if you want a 4096 bit key use 'ipfs init -b 4096'
<drathir>
whyrusleeping: oh thanks a lot thats great...
<lgierth>
drathir: could you fix the doc saying that?
<lgierth>
and ohai btw :)
<lgierth>
(one more hyperborian)
joshbuddy has joined #ipfs
<drathir>
lgierth: hi... yea goin to find and fix that ^^
border has joined #ipfs
<drathir>
ladekjaer: and yea hype onboard... and honestly that from i hear about project...
voxelot has quit [Ping timeout: 246 seconds]
<ralphtheninja>
ansuz_: ping
joshbuddy has quit [Quit: joshbuddy]
<drathir>
also not much time ago one of ipfs dev team member also say "few" good words about project that why decided to try by myself how that workin... gz for presentation... both of them ccc and website one...
<spikebike>
gz = good zebra?
<whyrusleeping>
gratz bruh
border has quit [Read error: Connection reset by peer]
<spikebike>
ah
<drathir>
lgierth: done...
<drathir>
spikebike: gz=congratulations
<lgierth>
drathir: cool! :)
<lgierth>
which repo was it? looks like i missed to enable notifications for this particular one
<jbenet>
lgierth: have you tried pushing out an update to ipfs/website before?
<jbenet>
its a bit convoluted because gh-pages + pinning, and so on. but things will be easier when we have (a) ipfs.io working out of there, and (b) a github pinbot.
<jbenet>
we should totally make a "pin-to-ipfs" github webhook!
<lgierth>
yeah, heh
<lgierth>
i have built the website locally, pinned it and used that for ipfs.io's TXT record
disusered has joined #ipfs
<whyrusleeping>
lgierth: would be awesome if we had a dashboard to see all our gateway nodes storage levels
<lgierth>
whyrusleeping: yeah i've been meaning to have that at metrics.i.ipfs.io
border has joined #ipfs
patcon has quit [Ping timeout: 245 seconds]
nessence has quit []
disusered has quit [Ping timeout: 256 seconds]
border has quit [Read error: Connection reset by peer]
notduncansmith has quit [Read error: Connection reset by peer]
sharky has quit [Ping timeout: 240 seconds]
sharky has joined #ipfs
notduncansmith has joined #ipfs
notduncansmith has quit [Read error: Connection reset by peer]
pfraze has quit [Remote host closed the connection]
notduncansmith has joined #ipfs
notduncansmith has quit [Read error: Connection reset by peer]
notduncansmith has joined #ipfs
notduncansmith has quit [Read error: Connection reset by peer]
caseorga_ has joined #ipfs
notduncansmith has joined #ipfs
notduncansmith has quit [Read error: Connection reset by peer]
Leer10 has joined #ipfs
notduncansmith has joined #ipfs
notduncansmith has quit [Read error: Connection reset by peer]
notduncansmith has joined #ipfs
notduncansmith has quit [Read error: Connection reset by peer]
caseorga_ has quit [Remote host closed the connection]
dignifiedquire has joined #ipfs
akhavr has quit [Ping timeout: 246 seconds]
dignifiedquire has quit [Client Quit]
akhavr has joined #ipfs
simonv3 has quit [Quit: Connection closed for inactivity]
notduncansmith has joined #ipfs
notduncansmith has quit [Read error: Connection reset by peer]
slothbag has joined #ipfs
caseorga_ has joined #ipfs
joshbuddy has quit [Quit: joshbuddy]
Leer10 has quit [Ping timeout: 272 seconds]
caseorga_ has quit [Ping timeout: 246 seconds]
slothbag has quit [Remote host closed the connection]
bedeho_ has quit [Ping timeout: 245 seconds]
bedeho has quit [Ping timeout: 256 seconds]
notduncansmith has joined #ipfs
notduncansmith has quit [Read error: Connection reset by peer]
compleatang has joined #ipfs
inconshr_ has quit [Ping timeout: 244 seconds]
inconshreveable has joined #ipfs
notduncansmith has joined #ipfs
notduncansmith has quit [Read error: Connection reset by peer]
notduncansmith has joined #ipfs
notduncansmith has quit [Read error: Connection reset by peer]
Leer10 has joined #ipfs
notduncansmith has joined #ipfs
notduncansmith has quit [Read error: Connection reset by peer]
notduncansmith has joined #ipfs
notduncansmith has quit [Read error: Connection reset by peer]
wiretapp1d-cb is now known as wiretapped-cb
notduncansmith has joined #ipfs
notduncansmith has quit [Read error: Connection reset by peer]
<cryptix>
gmorning ipfs
Encrypt has joined #ipfs
<M-davidar>
morning cryptix
<M-davidar>
how's the git on ipfs stuff going?
slothbag has joined #ipfs
<cryptix>
mmmh... delayed due to a cold this week. did some housekeeping on the code (tests now run on travis)
<cryptix>
still blocking on push and figuring out the process.. need some enlightenment on the documentation..
<cryptix>
i wanted to reuse how git does things with ssh, but that remote helper does a lot of the magic itself (calls out to ssh) - i hoped i could just copy that and strip out the ssh but... the tool layers are a mess imho
<cryptix>
M-davidar: im working on it today. maybe it will finally make 'click'
dignifiedquire has joined #ipfs
<M-davidar>
cryptix: would reusing the http remote be easier than the ssh one?
<cryptix>
well.. from a practical standpoint: you can work today with just whyrusleeping git-ipfs-rehost tool
akhavr1 has joined #ipfs
<cryptix>
you can publish your repo with it and pull from http://gateway...
<cryptix>
but that isnt documented for push. the smart protocol on the other hand expects another process working on another copy of the repo - i hoped to bypass the step of cloning the remote locally and working on that one
<cryptix>
but it looks like i need to go that way again until I understand the process better
notduncansmith has joined #ipfs
notduncansmith has quit [Read error: Connection reset by peer]
<M-davidar>
cryptix: how does git-remote-dropbox handle it?
<cryptix>
M-davidar: interesting, havent seend that one. thx
ei-slackbot-ipfs has quit [Remote host closed the connection]
ei-slackbot-ipfs has joined #ipfs
dignifiedquire has quit [Quit: dignifiedquire]
notduncansmith has joined #ipfs
notduncansmith has quit [Read error: Connection reset by peer]
marianoguerra has joined #ipfs
Encrypt has quit [Quit: Quitte]
<M-davidar>
anyone used hubiC?
<M-davidar>
10TB for 50eur/yr seems too good to be true...
<M-davidar>
(trying to work out if it would make a cheap but likely low-reliability ipfs blockstore)
dignifiedquire has joined #ipfs
<zorun>
M-davidar: apparently, transfer speeds are very slow
<zorun>
(don't remember in which direction)
caseorga_ has joined #ipfs
<M-davidar>
zorun: what about data integrity?
notduncansmith has joined #ipfs
notduncansmith has quit [Read error: Connection reset by peer]
<cryptix>
M-davidar: the git-remote-dropbox helps a lot - thanks
<cryptix>
i looked for others but missed that one
<cryptix>
they just use their CRUDesque api - no remote process involved - exactly what i wanted
<M-davidar>
cryptix: no worries, one of the benefits of a hn addiction :)
<cryptix>
:))
<cryptix>
plus: its in python! (meaning readable src)
caseorga_ has quit [Ping timeout: 244 seconds]
<M-davidar>
yeah, a lot of git's in perl isn't it?
<M-davidar>
i used to like perl
<cryptix>
i've seen sh, c, perl, python too irrc
<M-davidar>
then i came back to code I'd written a month earlier
<M-davidar>
didn't like it so much after that :)
<cryptix>
heheh - yea, .pl also didnt click for me
<M-davidar>
too many magic symbols
richardsmith has joined #ipfs
marianoguerra has quit [Ping timeout: 260 seconds]
<mg__>
I have quite a lot of bandwidth to spare, is there anything to pin that would help the project in any way?
<mg__>
cryptix: sweet, thanks :) I have access to a fairly underutilized 1gbit connection, and have atleast 250gb of space available
<cryptix>
nice! where is it located?
<mg__>
Norway
<cryptix>
cool
<cryptix>
the ipfs.io site is also moving to be hosted by the gateways soon so there is more to come :)
<mg__>
I presume stuff will be added to the file you linked?
<cryptix>
yup - that repo will be updated
<cryptix>
the seed.sh also pulls them and adds them to a local node
<mg__>
this is my first playing with ipfs so excuse the stupid questions! How can I set where the files will be physically located? I'd like to keep it on a scratch disk
notduncansmith has joined #ipfs
notduncansmith has quit [Read error: Connection reset by peer]
dignifiedquire has quit [Quit: dignifiedquire]
dignifiedquire has joined #ipfs
dignifiedquire has quit [Client Quit]
<cryptix>
mg__: np :) it's all inside $HOME/.ipfs by default, you can either move that hole repo to another location by exporting IPFS_PATH (good for multiple nodes on a system)
<cryptix>
or change the Datastore.Path location in $IPFS_PATH/config
<cryptix>
wait.. no.. datastore.path is the index
<cryptix>
im not sure there is a config option to point IPFS_PATH/blocks somewhere else (cc jbenet whyrusleeping)
<mg__>
hm, ok. So the easiest is just to mv the .ipfs folder to the wanted location, and export IPFS_PATH in .bashrc?
inconshreveable has quit [Read error: Connection reset by peer]
<cryptix>
yup
inconshreveable has joined #ipfs
<mg__>
cryptix: ok. The paths in the config are full paths though, so I presume I have to change those too
<cryptix>
mg__: not sure if that is relative to IPFS_PATH by itself - to be sure, yea make them full paths
<M-davidar>
would symlinking work?
<mg__>
hm, the seed.sh command hangs on ipfs refs for quite a while
<cryptix>
mg__: yup - thats the pulling process
<cryptix>
M-davidar: not sure about symlink - maybe, try it :)
<cryptix>
i assume it would
notduncansmith has joined #ipfs
notduncansmith has quit [Read error: Connection reset by peer]
<M-davidar>
lol, seed.sh OOM'd the daemon
<M-davidar>
ipfs really isn't kind to small vps's :)
<M-davidar>
cryptix: adding -L1 to xargs in seed.sh makes it a bit nicer
<cryptix>
M-davidar: open to PRs :)
<M-davidar>
ok, will do
<M-davidar>
done
notduncansmith has joined #ipfs
notduncansmith has quit [Read error: Connection reset by peer]
notduncansmith has joined #ipfs
notduncansmith has quit [Read error: Connection reset by peer]
<M-davidar>
cryptix: I'm stuck on Qmch7G2F4MqT22z1f6uYnc2rAQZGD5p7xXsnsJnSYyN2Q7
<M-davidar>
public gateway doesn't seem to be able to find it either
slothbag has quit [Remote host closed the connection]
notduncansmith has joined #ipfs
notduncansmith has quit [Read error: Connection reset by peer]
Encrypt has joined #ipfs
<cryptix>
M-davidar: hrm - me neither.. can you tell me where that is from?
<cryptix>
which block of refs at least
<M-davidar>
cryptix: one sec
<M-davidar>
cryptix: seems to be referenced by QmQVrrQQxXjf3GESdQc6ajF6TVoNpkszAMMcJLJJczoRJz
notduncansmith has joined #ipfs
notduncansmith has quit [Read error: Connection reset by peer]
<mg__>
So how long should a full seed.sh take? my repo doesn't grow (it's only 13M), and the seed.sh script has been stuck for 30 mins, last thing to be printed was QmUzorLM61MJzKaPtX2jGceKuYBdeXitgNim4AjdrMYGBw which seems to be the ipfs logo
<cryptix>
jea.. looks like not all of it is properly seeded.. M-davidar also noticed :-/ cc jbenet
<mg__>
I see.. there's no timeout?
<cryptix>
not on, 'ipfs refs' - we should add a flag to enable timout. 'ipfs get/cat' do
<M-davidar>
cryptix: the problem seems to be the files in /ipfs/QmQVrrQQxXjf3GESdQc6ajF6TVoNpkszAMMcJLJJczoRJz/api/demoapp/
<cryptix>
oh well
notduncansmith has joined #ipfs
notduncansmith has quit [Read error: Connection reset by peer]
<cryptix>
it also looks like the ipfs.io/docs/examples links don't point to the examples hash listed in infra/.../refs-to-seed - just ran make in https://github.com/ipfs/examples/tree/master/examples and it gave me a different root hash (QmPoT57fd6gcAmH7PM5PKzNkVu8K52gXi9dSz8JM4UwXvG)
<cryptix>
the 'published-version' in that repo is also 3month old...
<cryptix>
cc lgierth jbenet
<cryptix>
QmQVrrQQxXjf3GESdQc6ajF6TVoNpkszAMMcJLJJczoRJz < which is in infra/../refs-to-seed is also 2 commits behind
richardsmith has quit [Quit: richardsmith]
richardsmith has joined #ipfs
richardsmith has left #ipfs [#ipfs]
notduncansmith has joined #ipfs
notduncansmith has quit [Read error: Connection reset by peer]
inconshreveable has quit [Remote host closed the connection]
_whitelogger_ has joined #ipfs
_whitelogger has quit [Ping timeout: 240 seconds]
daviddias has quit [Ping timeout: 240 seconds]
jbenet has quit [Ping timeout: 240 seconds]
ZioFork has quit [Ping timeout: 240 seconds]
tibor has quit [Ping timeout: 240 seconds]
ebarch has quit [Ping timeout: 240 seconds]
otherchas__ has quit [Ping timeout: 240 seconds]
dandroid has quit [Ping timeout: 240 seconds]
sindresorhus has quit [Ping timeout: 240 seconds]
celehner has quit [Ping timeout: 240 seconds]
okket has quit [Ping timeout: 240 seconds]
sbruce has quit [Ping timeout: 240 seconds]
luca has quit [Ping timeout: 240 seconds]
luca has joined #ipfs
dandroid_ is now known as dandroid
daviddias_ is now known as daviddias
<cryptix>
looks like irccloud is having problems today? o.o
sindresorhus_ is now known as sindresorhus
otherchas___ is now known as otherchas__
ebarch has joined #ipfs
ZioFork has joined #ipfs
notduncansmith has joined #ipfs
notduncansmith has quit [Read error: Connection reset by peer]
celehner has joined #ipfs
dignifiedquire has joined #ipfs
okket_ has quit [Remote host closed the connection]
jbenet has joined #ipfs
okket has joined #ipfs
dignifiedquire has quit [Client Quit]
dignifiedquire has joined #ipfs
dignifiedquire has quit [Client Quit]
notduncansmith has joined #ipfs
notduncansmith has quit [Read error: Connection reset by peer]
notduncansmith has joined #ipfs
notduncansmith has quit [Read error: Connection reset by peer]
cSmith has quit [Changing host]
cSmith has joined #ipfs
ansuz_ is now known as ansuz
<drathir>
any idea what can cause package github.com/ipfs/go-ipfs/cmd/ipfs: unable to parse output of git config remote.origin.url
notduncansmith has joined #ipfs
notduncansmith has quit [Read error: Connection reset by peer]
<cryptix>
drathir: uhm.. from what command to you get that?
<drathir>
go get -u github.com/ipfs/go-ipfs/cmd/ipfs
<drathir>
not found yet the tutorial of using git clonned version instead...
<cryptix>
interesting... can you go to $GOPATH/src/github.com/ipfs/go-ipfs and show me the output of 'git remote -v' ?
<drathir>
maybe i will try remove and download again...
<drathir>
btw its 1:1 normal git clone?
<cryptix>
what golang version do you have installed? 'go version'
<drathir>
go version go1.5 linux/amd64
<cryptix>
yea.. usually 'go get' just shells out to git pull afaik
<cryptix>
cant reproduce.. if you dont have unpublished branches in that git repo, can you just delete it and try again? 'go get github.com/ipfs/go-ipfs/cmd/ipfs'
<drathir>
yea its clean install...
<drathir>
wonder yesterday on another box go w/o problem... ^^
<cryptix>
i just did a fresh gopath, get worked and get -u also does
<drathir>
cryptix: thanks a lot that mean at my side need search..
<cryptix>
did you clone go-ipfs manually?
chriscool has joined #ipfs
<drathir>
cryptix: no, but i will check that too...
<drathir>
that box is little overloaded maybe proces interrupted or somethin like that, probably need to reboot soon...
<drathir>
also need to search, bc i woif peers workin in way user chose with who want peer of it randomly peering with ppl sharing the ip, and what happen in case nat is used...
<drathir>
wonder if*
notduncansmith has joined #ipfs
notduncansmith has quit [Read error: Connection reset by peer]
JasonWoof has quit [Remote host closed the connection]
<bret>
yeah, and it can be run off peoples university shell accounts :p
Eudaimonstro has quit [Read error: No route to host]
Eudaimonstro has joined #ipfs
voxelot has joined #ipfs
Eudaimonstro has quit [Ping timeout: 246 seconds]
Eudaimonstro has joined #ipfs
pfraze has quit [Remote host closed the connection]
dignifiedquire has joined #ipfs
domanic has quit [Read error: Connection reset by peer]
notduncansmith has joined #ipfs
notduncansmith has quit [Read error: Connection reset by peer]
inconshreveable has joined #ipfs
Eudaimonstro has quit [Ping timeout: 245 seconds]
pfraze has joined #ipfs
dignifiedquire has quit [Quit: dignifiedquire]
notduncansmith has joined #ipfs
notduncansmith has quit [Read error: Connection reset by peer]
Eudaimonstro has joined #ipfs
Eudaimonstro has quit [Client Quit]
marianoguerra has joined #ipfs
JasonWoof has quit [Ping timeout: 240 seconds]
JasonWoof has joined #ipfs
jatb has quit [Remote host closed the connection]
JasonWoof has quit [Remote host closed the connection]
jatb has joined #ipfs
<drathir>
btw any ipfs address which workin to test connection?
JasonWoof has joined #ipfs
JasonWoof has joined #ipfs
<jbenet>
hey lgierth -- krl needs the CORS stuff for the gateways, can haz fix + deploy when you get a chance?
caseorga_ has joined #ipfs
Encrypt has joined #ipfs
JasonWoof has quit [Ping timeout: 245 seconds]
JasonWoof has joined #ipfs
chriscool has quit [Ping timeout: 256 seconds]
JasonWoof has quit [Ping timeout: 260 seconds]
JasonWoof has joined #ipfs
JasonWoof has joined #ipfs
caseorga_ has quit [Remote host closed the connection]
chriscool has joined #ipfs
<ipfsbot>
[go-ipfs] jbenet created check-for-api (+1 new commit): http://git.io/vsV6E
<ipfsbot>
go-ipfs/check-for-api 0fe6fe2 Juan Batiz-Benet: check for API -- WIP...
<bret>
drathir what do you mean?
<bret>
hashes?
<whyrusleeping>
jbenet: lol, its kindof complicated
<whyrusleeping>
i'm not sure though that i want to separate the daemon from the client entirely
dignifiedquire has joined #ipfs
<whyrusleeping>
its kinda nice having just one command
<ehmry>
I say make a dedicated ipfsd
<whyrusleeping>
ehmry: hrm... but then you have to install two separate binaries
<ehmry>
but that doesn't mean abandoning ipfs
<ehmry>
yes
<whyrusleeping>
i mean, if people dont mind that then i'm okay with it
<ehmry>
I would like to have a system daemon that ran under special permissions and started at boot
<ehmry>
but I'm just a low level guy
<voxelot>
I'd like too ehmry for my web application, so users don't have to even know ipfs is running when they interact with the site
<whyrusleeping>
ehmry: so workflow would be running 'ipfs init' and then 'ipfsd'
<whyrusleeping>
or maybe ipfsd should be able to initialize the repo on its own?
<ehmry>
yea, thats what I was thinking
<jbenet>
whyrusleeping we could do the "launch ipfsd in the background with refcounts" if ipfsd is not on.
<jbenet>
whyrusleeping and forgo doing things in a single process.
<jbenet>
(not sure still if this is _the best_ way of doing things, but it is _a_ way that makes other things simpler.
<ehmry>
I would think this would mean duplicating some functionality between clients and the daemon but the deamon would be handling storage and networking
caseorga_ has joined #ipfs
<jbenet>
yep. the thing we lose is "git-like" functionality of manipulating the repo without an online service. which is typically much faster. (i sometimes kill my daemon to add huge things)
<jbenet>
tough decisions.
caseorga_ has quit [Remote host closed the connection]
<drathir>
bret: somethin to check connection if workin...
<drathir>
btw ipfs support tcp of udp connections?
<whyrusleeping>
drathir: ipfs uses tcp right now
<whyrusleeping>
but we want to support a udp based protocol soon
<drathir>
whyrusleeping: thats good to have tcp choice...
<whyrusleeping>
yeah, tcp is less likely to be blocked
<drathir>
whyrusleeping: user choice option could be possible eg. only tcp/ only udp/ both ?
marianoguerra has quit [Ping timeout: 272 seconds]
pfraze has quit [Remote host closed the connection]
<whyrusleeping>
yeah, youll be able to pick what protocol you want your node to use for its listener port
<whyrusleeping>
just by modifying the config
<jbenet>
(and both)
<drathir>
udp will be good for media services i guess...
domanic has joined #ipfs
reit has quit [Ping timeout: 240 seconds]
<bret>
drathir: here is something hosted off my raspi at home QmdY53gvcuKQgyRQbg1n9LJMBJvcdvpjdGHuMP53LTu3kF
<jbenet>
batch 1 is all the ipld stuff. which is super close
<jbenet>
batch 2 is the records + keystore. records are specced, just need batch 1 in. keystore needs a bit more specing, but we have a strong start
<jbenet>
batch 3 is the dht naming stuff + migration, which come last.
<whyrusleeping>
yeap
<jbenet>
i claim batch 1 can be finished this week if we focus on landing it.
<jbenet>
which would be a strong push
<jbenet>
then thers 4, which are close, right?
<whyrusleeping>
for batch one, we need the fixes to the cbor encoder?
<whyrusleeping>
4?
<jbenet>
yeah we do
<jbenet>
4 is below the line
<whyrusleeping>
ah, gotcha
<whyrusleeping>
those are mostly ready to land
<whyrusleeping>
just need some CR, and a small amount of debugging on one
<jbenet>
ok, i'll CR tonight
<jbenet>
so, batch 5
<jbenet>
batch 5 is important too
<jbenet>
so are 8 and 9.
<jbenet>
8 more than 9 i think
<whyrusleeping>
yeah, 8 seems important
<whyrusleeping>
the memory issue is one thats been around for a while now
<whyrusleeping>
although i just thought it was the GC being lazy as i had never seen it crash before
<jbenet>
i think it's probably a mistake in our adding that prevents the GC from deallocating things until the entire request ends
<jbenet>
(like, we hang on to ptrs that GC cant get rid of yet.
<jbenet>
cause afaict the memory climb happens only in one single add
<Bat`O>
what would be needed to see the "tracking datastore" moving forward ?
<whyrusleeping>
jbenet: okay, that shouldnt be too hard to track down
<Bat`O>
sorry to interrupt, hi :)
<whyrusleeping>
Bat`O: someone to spec it out probably
<jbenet>
Bat`O probably a sample implementation? try giving it a shot. or maybe whyrusleeping can, but i think he's got a ton of things that are very pressing already
<Bat`O>
i don't think i know enough ipfs internal and go to do that :-\
<domanic>
jbenet, whyrusleeping hey what signing algorithm is used in ipfs?
<jbenet>
whyrusleeping rht__ ok let's work on: batches: 1, 4, 7, 8 this week. think could do all that?
<Bat`O>
i guess i could help to implement is someone knowledgable would define a battle plan
<jbenet>
domanic: multikey.
<domanic>
jbenet, what signing algorithm do you use by default/rightnow
<jbenet>
domanic: none. we dont have object signing baked in yet. it's coming right up un batch 2 though
<whyrusleeping>
domanic: RSA is used right now for the few things we sign
<whyrusleeping>
PKCS1v15 is the exact one that may be in use
<domanic>
whyrusleeping, dude, use sodium & ed25519
<whyrusleeping>
domanic: we want to, it just needs to be integrated into the codebase
<whyrusleeping>
and we havent gotten to that yet
<jbenet>
(oh yeah we do sign things, that's right)
<jbenet>
domanic: PRs welcome :)
<whyrusleeping>
jbenet: "move along, theres no crypto here"
bedeho has quit [Quit: Leaving]
<jbenet>
whyrusleeping: i'll CR 4 and 7 tonight.
<whyrusleeping>
jbenet: sgtm
<jbenet>
whrysleeping if you can do the cbor thing, i can do the protobuf thing. the only large thing left in 1 is pub/priv
<jbenet>
whyrusleeping what state is that in?
<whyrusleeping>
i think it needs CR?
<jbenet>
ok i'll CR that too.
<whyrusleeping>
let me check
<whyrusleeping>
yeah, it needs discussion/CR
<jbenet>
ok
<jbenet>
want to order which PRs i should CR first?
<whyrusleeping>
jbenet: i changed the order on the etherpad
<jbenet>
ok thanks
<jbenet>
so, let's try and land 1, 4, 7, 8 this week. then 2, 5, then 3, 9, then 6
<jbenet>
everyone willing the help us with all this, would be huge. the faster we get through it all, the faster we get robust IPNS!
<whyrusleeping>
apparently the entire state of washington as well as most of idaho is on fire
<whyrusleeping>
theres a dangerous air conditions warning out due to too much smoke
<jbenet>
moats are useufl
<mg__>
jbenet: where's the batches described?
<whyrusleeping>
moats keep the fire out, but not the smoke
notduncansmith has joined #ipfs
notduncansmith has quit [Read error: Connection reset by peer]
<mg__>
jbenet: thanks :) I'd like to start contributing, but I'd have to familiarize myself with the codebase a bit first I think. Handing in my thesis in 48 hours though, will get on it after that :)
JasonWoof has quit [Remote host closed the connection]
JasonWoof has joined #ipfs
<jbenet>
mg__ awesome! congrats, what's it on?
<mg__>
jbenet: thanks! active selection in deep neural networks (basically how to choose training examples that are especially likely to be useful, to avoid unnecessary labeling)
<jbenet>
mg__ cool! have links to your work? would love to check it out! (i like ML quite a bit)
notduncansmith has joined #ipfs
notduncansmith has quit [Read error: Connection reset by peer]
Eudaimonstro has joined #ipfs
<jbenet>
(also, there are many applications of ML to ipfs-things, we may be able to work on some very novel work there
<jbenet>
blame: didn't you make python bindings for the ipfs daemon?
<whyrusleeping>
jbenet: i know he did a python multihash
<whyrusleeping>
but i dont remember api bindings
<whyrusleeping>
(at least, they arent on his github)
<voxelot>
I'm checking out SVM myself mg_ , interested in understanding the advantages of finding global minima over the ANN or DNN local minima
<mg__>
jbenet: no papers out yet, but I can send you a link to the thesis in 48 hours :p Working on a paper on it too, but that'll take some more time. The parts of the thesis that are interesting are short anyway.
<mg__>
voxelot: interesting, how would that work?
<voxelot>
check out support vector machines
<jbenet>
mg__ sweet, please do
<mg__>
voxelot: Yeah I know SVMs fairly well, but I didn't quite understand the link
<mg__>
voxelot: or rather, how SVMs can find global minima over DNN local minima. But it sounds interesting!
<voxelot>
ohh right :) well if you are using multilayered perceptrons then they are in direct competition with svm's from what i hear. I just think it's interesting to see the research expanding on ANN's to compete
<mg__>
jbenet: did you have anything specific in mind, re ipfs and ML? Obviously lots of stuff that can be looked at, but would be interested to hear what you're thinking as someone that has thought more about ipfs than most (I presume!)
pfraze has quit [Remote host closed the connection]
<voxelot>
maybe ML to improve network efficency?
<voxelot>
finding correct PNS in DHT
Eudaimonstro has quit [Quit: Leaving]
<whyrusleeping>
jbenet: i fixed the issue you were having with gx
<whyrusleeping>
and added a version command
<whyrusleeping>
i'm also thinking of renaming it to gex
<whyrusleeping>
since thats how i pronounce it
JasonWoof has quit [Ping timeout: 260 seconds]
<whyrusleeping>
oh, and now when you import a package, it symlinks it under the vendor dir so you can have nice import names
pfraze has joined #ipfs
Encrypt has quit [Quit: Sleeping time!]
JasonWoof has joined #ipfs
inconshreveable has quit [Ping timeout: 244 seconds]
notduncansmith has joined #ipfs
notduncansmith has quit [Read error: Connection reset by peer]
<jbenet>
mg__ many things, from selection connections, bitswap sending, and even (once we get to this) network topology. there's a _massive_ amount of work to be done mixing ML successes with the network. TCP Remy is just the first of a long string of important results coming.
<jbenet>
mg__ (as far as ML is concerned, protocol development is in the dark ages)
<whyrusleeping>
with this, i can now click ipfs links in my terminal :)
<mg__>
jbenet: Remy is interesting for sure. And yeah I agree, ML right now is surfing on a wave (me included) of deep learning image/speech/etc-successes, will be interesting to see where stuff turns when the last drop of performance is squeezed out of the benchmarks and people start looking elsewhere
<mg__>
Like, can we actually apply the stuff to anything other than detecting cats in images
<mg__>
Of course, I think many of the more interesting applications in networking are probably more the domain of statistical methods more than sub-symbolic neural stuff
<voxelot_>
you want more than cats?? :(
<voxelot_>
mg__ you saw googles chat bot in july right?
<mg__>
voxelot_: yeah google puts out a lot of cool stuff on the ML front. Same with facebook, baidu etc
<jbenet>
mg__ yeah. super exciting. i took ML from Ng in 2011, so i caught a lot of the "generalized learning alg" excitment, like the sparse autoencoders work, and so on. actually worked with Ng on reproducing some of Bengio's CNN results.
<mg__>
cool! he's at baidu now, right?
<jbenet>
last i heard yeah
<jbenet>
(i think the best team atm is google brain though-- Geoff Hinton + Jeff Dean is the only such pairing possible.
<mg__>
haha yeah, hinton is a machine
zabirauf has quit [Quit: My Mac has gone to sleep. ZZZzzz…]
<jbenet>
brb
<mg__>
Tbh I'm becoming a bit disillusioned with the supervised DL craze though. I'm more interested in stuff like what deepmind are doing with the DQNs
<lgierth>
jbenet: i dont have a chance to fix it right now
<lgierth>
jbenet: krl: mappum: come to bethanien if you fancy a party
<alu>
does ipfs mess up with VPN's?
<lgierth>
i'm organizing it so i can't really get out for an hour and deploy stuff
<lgierth>
i can do it before going to bed, ETA 7 am
<lgierth>
jerome and ansuz are here too
<lgierth>
bbl
notduncansmith has joined #ipfs
notduncansmith has quit [Read error: Connection reset by peer]
patcon has quit [Ping timeout: 246 seconds]
inconshreveable has joined #ipfs
notduncansmith has joined #ipfs
notduncansmith has quit [Read error: Connection reset by peer]
inconshreveable has quit [Ping timeout: 246 seconds]
patcon has joined #ipfs
caseorga_ has joined #ipfs
caseorga_ has quit [Ping timeout: 244 seconds]
Leer10 has quit [Remote host closed the connection]
<jbenet>
lgierth: yep no worries at all!
<jbenet>
mg__: yeah deep mind is very exciting
notduncansmith has joined #ipfs
notduncansmith has quit [Read error: Connection reset by peer]
JasonWoof has quit [Ping timeout: 264 seconds]
<mg__>
jbenet: yeah, I think reinforcement learning is a cool field, with potentially more interesting applications than straight supervised stuff. Will be interesting once they can get to the point where the networks are not purely reactionary
<mg__>
I assume they're working on something like that now, and win that montezuma game they sucked at in the nature paper :p
reit has joined #ipfs
<whyrusleeping>
one prof at my university was working on having agents teach each other how to play games
<mg__>
what uni?
<whyrusleeping>
washington state
<whyrusleeping>
prof is matthew taylor
<mg__>
cool
<mg__>
whyrusleeping: do you have an AI-background too?
<whyrusleeping>
not super heavy, but i've done a good deal
<whyrusleeping>
neural nets and machine learning stuff to try and detect network attacks
<mg__>
I think its good to have a certain feel for it for all CS-graduates nowadays. Can come in handy in many situations
<mg__>
aha
<whyrusleeping>
yeah, everything is AI
<whyrusleeping>
well, until it works that is
<mg__>
haha, that's right
notduncansmith has joined #ipfs
notduncansmith has quit [Read error: Connection reset by peer]
<voxelot_>
AI is just cool, think it attracts all computer scientists
<voxelot_>
and for the cryptographers out there, homomorphic encryption is reall cool
<voxelot_>
just found this article today about some bitcoin guys and MIT working on it