<ipfsbot>
[go-ipfs] whyrusleeping created version-bump-0.3.8 (+1 new commit): http://git.io/vC39a
<ipfsbot>
go-ipfs/version-bump-0.3.8 6ad4015 Jeromy: ipfs version 0.3.8 changelog and version bump...
domanic has quit [Ping timeout: 264 seconds]
<ipfsbot>
[go-ipfs] whyrusleeping opened pull request #1822: ipfs version 0.3.8 changelog and version bump (master...version-bump-0.3.8) http://git.io/vC3H8
qgnox has left #ipfs ["WeeChat 1.3"]
<achin>
oooooh
<M-rschulman1>
exciting times
<nikogonzo>
interesting, symlinks. what was the use case for that?
Not_ has quit [Ping timeout: 246 seconds]
<achin>
linking to things symbolicly!
<ipfsbot>
[go-ipfs] ForrestWeston opened pull request #1823: Pin commands default to recursive (master...recurPin) http://git.io/vC37s
amstocker has quit [Ping timeout: 244 seconds]
<whyrusleeping>
nikogonzo: 'ipfs add -r' can properly handle symlinks
<whyrusleeping>
previously it would follow the symlink, and add that file
<whyrusleeping>
which meant that it would break on broken symlinks
<nikogonzo>
cool; how will it handle recursive symlinks?
<whyrusleeping>
it uses the os'es symlink resolution logic
<whyrusleeping>
which for linux means it follows at most 32 symlinks
<nikogonzo>
neat, i wil derploy the new version in a little bit
<nikogonzo>
thanks :)
<whyrusleeping>
nikogonzo: oh, its not out yet
<whyrusleeping>
thats the staging for the version cut
<nikogonzo>
oh gotcha
vrs has left #ipfs ["WeeChat 1.2"]
DrPete has joined #ipfs
DrPete_ has quit [Ping timeout: 265 seconds]
SWingedSeraph_ has joined #ipfs
SWingedSeraph has quit [Ping timeout: 272 seconds]
<sonatagreen>
i mostly want it to handle symlinks to /ipfs/hashxyz correctly
<sonatagreen>
and if you 'add -r foo' then it should handle a link from foo/bar to foo/baz
<whyrusleeping>
yeap, it can do that
hellertime has joined #ipfs
<whyrusleeping>
it adds it as an actual symlink
<davidar>
whyrusleeping (IRC): will the gateway also follow symlinks properly?
<whyrusleeping>
davidar: ehhhhhhh, wiring symlinks into the gateway resolution gets tricky
<whyrusleeping>
so i dont beleive it will resolve it
<davidar>
hrm
<davidar>
whyrusleeping (IRC): will there be an option to just follow symlinks when adding then?
voxelot has joined #ipfs
<davidar>
whyrusleeping (IRC): the gateway could just trigger redirects for symlinks, no?
<davidar>
*http redirects
<davidar>
Actually, that might not work well
<davidar>
Internal redirects, like how ipns works currently
<whyrusleeping>
yeah, i think add has an option to follow symlinks
<davidar>
Cool
<whyrusleeping>
or maybe not
<whyrusleeping>
i lied
<davidar>
hrm
<whyrusleeping>
yeap, no such option...
<whyrusleeping>
hemmm
<davidar>
I kind of figured dedup rendered symlinks obsolete anyway outside of ipns
<whyrusleeping>
it does
<achin>
except when you "ipfs get"
<whyrusleeping>
^
<whyrusleeping>
and for things like docker
<whyrusleeping>
we wanted the ability to run a docker rootfs out of ipfs
<whyrusleeping>
and docker rootfs's require symlinks
<davidar>
IMO literal symlinks shouldn't be the default until the gateway supports them
<davidar>
Otherwise we end up with a situation like shortcuts on windows :/
<davidar>
s/gateway/path resolution in general
<multivac>
davidar meant to say: IMO literal symlinks shouldn't be the default until the path resolution in general supports them
<gaboose>
ive decided to work on a ipfs-friendly distributed multicast system (pub-sub style) that would enable push notifications. not using a federation of trust decentralized like Matrix, and using MST topology, not gossip like Whisper
<Gaboose_>
do you guys think you need something like that?
<Gaboose_>
i meant, not using a trusted federation like Matrix
<Gaboose_>
written in Go and all
<Gaboose_>
jbenet whyrusleeping
amstocker has quit [Ping timeout: 246 seconds]
<spikebike>
Gaboose_: I've been pondering similar
<spikebike>
basically a message layer for messages between IPFS nodes that would allow routing via DHT, and store/forward if the target isn't online
tsenart has quit [Quit: My Mac has gone to sleep. ZZZzzz…]
<Gaboose_>
the main differences to that would be, no store/forward if the target isn't online and theoretically lower latency, because routing through MST and not DHT
<Gaboose_>
and messages are addressed to topics, not individual nodes
<Gaboose_>
so everyone on a website would subscribe to the topic 'twitter feed' for example
<Gaboose_>
"layers over IPRS to do discovery" what is IPRS?
<Gaboose_>
davidar, thanks for the info! a lot of discussion there happened 1 - 3 days ago. i hope i'm not too late to the party
<davidar>
Gaboose_ (IRC): not at all, still plenty of time to be suggesting ideas :)
<davidar>
Gaboose_ (IRC): nobody's actively working on pubsub yet
<davidar>
AFAIK, at least
<Gaboose_>
well I'm commited to it, it's my individual university project this year :)
<davidar>
Gaboose_ (IRC): awesome, comment on that issue and/or talk to jbenet about it
<Gaboose_>
will do
<davidar>
cc Matthew, Erik
<Gaboose_>
who's that?
<davidar>
Gaboose_ (IRC): matrix guys, I'm sure they'll be interested even if you're using a different model than federation
<ilyaigpetrov>
gaboose: I want true p2p chat, twitter and have some ideas I want to share with you and this channel. I propose a tagged messaging which may be more abstract than pub/sub: https://gist.github.com/ilyaigpetrov/02165e3ef4469a280344
<ilyaigpetrov>
What do you, guys, think of tagged messaging?
NeoTeo has quit [Quit: My Mac has gone to sleep. ZZZzzz…]
alu has quit [Quit: WeeChat 1.4-dev]
niek has quit [Ping timeout: 264 seconds]
<Gaboose_>
well, your specification can be reduced to get rid of setR by including #moscow-sell-house...(all tags concatenated to one tag and maybe hashed?) to setT
niekie has joined #ipfs
M-erikj is now known as M-erikj1
<Gaboose_>
ilyaigpetrov: i think tags (some people refer to them as topics or channels) are a reasonable extention to what ipfs guys are talking about when they mention pub-sub
<Gaboose_>
it's all about implementation at that point
<Gaboose_>
there are many different ways to build p2p topologies, routing strategies etc
M-erikj1 is now known as erikj`
<erikj`>
davidar: Thanks for the heads up. All: I'm happy to chat about messaging and give my perspective :) Though I need more coffee first...
<erikj`>
Gaboose_: I'm intrigued by the notion of no store and forward? If servers weren't reachable at the time, would you just drop those messages?
<ilyaigpetrov>
Gaboose_: With tags you can subscribe to {#ipfs #matrix}, {#ipfs} or {#matrix}, but with channels only the last two are possible
<Gaboose_>
hi erikj`! there would be no actual servers in that case, nodes would be all distributed, so yes, you would either drop those messages or store that last n-messages (fewer than is burdensome to client nodes) to replay to new subscribers
<Gaboose_>
permenant history storage could be done in cooperation with other technologies dedicated to storage. but this one would be dedicated to (ephemeral) messaging only
<Gaboose_>
just to focus on one thing at a time :)
<erikj`>
Fair enough, I would suggest having some sort of store and forward (even if bounded) as the internet has a tendency to break more often than you would think : (
<Gaboose_>
ilyaigpetrov: isn't subscribing to {#ipfs #matrix} the same as subscribing to {#ipfs} + {#matrix}?
<erikj`>
So these nodes/clients would basically be on desktops rather than, say, phones?
<ilyaigpetrov>
Gaboose_: no, you want messages about ipfs and matrix at the same time, it's AND, not OR
<davidar>
Gaboose_ (IRC): I suspect ilyaigpetrov is talking about subscribing to messages that contain both tags (not just one of them)
<davidar>
not sure how useful that would be though :/
<Gaboose_>
oh, so that's why i mentioned concatenating tags {#ipfs} + {#matrix} + {hash(ipfs-matrix)}
bedeho has quit [Ping timeout: 246 seconds]
<davidar>
Gaboose_ (IRC): yeah, I think that would make more sense
<Gaboose_>
erikj`: i can't see why it couldn't be done on phones too. would just need implement in another programming language etc
<davidar>
gaboose_ (IRC): but phones aren't online all the time
<Gaboose_>
erikj`: storing depends on the use case, i think. if you want to get notifications when new data is available, you don't need a history of all notifications
<ilyaigpetrov>
Gaboose_: I'm not sure why we need hashing at this stage of sketching, but suppose to emit a message to {#ipfs #matrix} I have to emit it to #ipfs, #matrix and #ipfs-matrix -- don't I?
<ilyaigpetrov>
oh, It's time to go for me, but I really want some feedback from you. Sorry, will be here later.
<Gaboose_>
thanks ilyaigpetrov, bye
<Gaboose_>
davidar: you mean you're concerned about message history? or...
<erikj`>
Gaboose: People tend to get upset if they miss notifications even when they think they're online. Agreed that you might not need to sync all missed notifications, but you'd still want to get the latest notification fairly reliably I would have thought
ilyaigpetrov has quit [Remote host closed the connection]
<erikj`>
but then I use IRC to obtain coffee at work in the mornings, and woe betide any messaging system that causes me to miss coffee :p
Encrypt has quit [Quit: Quitte]
<Gaboose_>
hm, right
<Gaboose_>
webapps that serve you infinite history usually do it piece by piece as you scroll upwards
<Gaboose_>
so that's already very different from messaging
<Gaboose_>
i guess there's different scenarios for no history vs n-latest history vs full history needs
<Gaboose_>
no history and n-latest history being possible with the "ephemeral" type of messaging
edt_ has quit [Quit: Leaving]
<Gaboose_>
and full history requires a storage system that's a whole different problem, i guess is my point :)
<erikj`>
I guess my point is that even when you have no history, you want some vague form of reliability. I can't really think of many scenarios where it'd OK to miss live notifications
<erikj`>
unless the notifications are sufficiently frequent
slothbag has quit [Quit: Leaving.]
gaboose has quit [Quit: Connection closed for inactivity]
<Gaboose_>
yea, i think i understand.. even with n-latest history, if the publisher shuts down for good, the subscriber will never get that message
<Gaboose_>
no servers - no reliability of that kind
<Gaboose_>
good point, fundumental flaw :)
<erikj`>
Yeah, servers make things *so* much easier :D. Depending on how funky you want to be, and the exact model you go for, you could probably put various mitigations in though
<erikj`>
anyway, I need to head off too, but this sounds quite interesting! I do have a soft spot for peer to peer messaging
<Gaboose_>
nice talking to you erikj`
<erikj`>
if you do a write up or something somewhere, i'd be interested in reading it :)
<davidar>
gaboose_ (IRC): tbh ephemeral messaging seems to conflict a bit with the whole permanent web idea
<ion>
erikj, Gaboose: Couldn't one provide the entire history as a chain of IPFS objects with an IPNS name and use pubsub just for instant update notifications? Someone who just got connected can receive what they missed through IPNS. Would the pubsub part need additional store and forward functionality in that case?
<davidar>
ion: yeah, I think it would make sense to leverage existing ipfs stuff as much as possible
<davidar>
then pubsub just needs to say "this thing has just updated"
<davidar>
(possibly with a copy of the update for efficiency)
besenwesen has quit [Quit: ☠]
tsenart has joined #ipfs
besenwesen has joined #ipfs
besenwesen has quit [Changing host]
besenwesen has joined #ipfs
<ilyaigpetrov>
Gaboose_: I'm here again, could you please comment: in pubsub to emit message to {#ipfs #matrix} you have to emit to #ipfs, #matrix and to #ipfs-matrix ?
<Gaboose_>
ion: i agree, but that would still need servers, which pin history though
Encrypt has quit [Quit: Quitte]
<Gaboose_>
davidar: i disagree, they don't have much in common, yes, but they do complement each other, pubsub needs storage and the web part of permenant web needs messaging
<erikj`>
Ion: Yup, that would work well when multiple nodes were subscribing to that topic/room or w.e. The problem is more to ensure the notifications themselves dont go awol. And you'd probably want to have history as a DAG rather than chain, at which point you'd probably want reliability for the last n notifications
<ion>
Any subscribers participating could automatically pin an amount of history they choose to retain.
<erikj`>
(Am on my phone ftr)
<zignig>
ion: Is it in a merkel dag or is this an external process ?
<zignig>
I have been thinking about it, the best i've found so far is to save a small text has and it timestamp.
<zignig>
write a series of known hashes into the DHT. nostril , nostril2015 , nostril201510 , nostril20151021 , etc
<zignig>
find the providers and the query their ipNs records, then make a block chain.
<Gaboose_>
ilyaigpetrov: yes, that's true. or more specifically setT(#ipfs #matrix), setR(#ipfs #matrix) -> setT(#ipfs #matrix #ipfs-matrix) and in another case setT(), setR(#ipfs #matrix) -> setT(#ipfs-matrix)
<ion>
zignig: I was thinking of the history as some append-only data structure on IPFS where an IPNS address is updated to point to the head. Pubsub is used at the minimum to send a pointer to the head but as davidar said, the message can include a copy of the new head object as well (which then has links to the history).
<zignig>
ion indeed , a pub-sub and/or a multicasting system would help. but a polling system is more durable.
* zignig
wants both. ;)
<davidar>
gaboose_ (IRC): you wouldn't *need* servers to pin history, but it would open up the possibility to do so
<davidar>
zignig (IRC): you can probably do polling now anyway
<zignig>
davidar: you can , did you see d97837b55e79595095a4b8f45732717f70586df1
<ion>
Subscribers can pin the latest object (which they received as a message or got through IPNS depending on whether they came online after the latest publish) and traverse forward through links to any degree they want to pin recent history.
<zignig>
lucky that wasn't a private gpg key.... ;/
<davidar>
gaboose_ (IRC): it doesn't really make sense to have two separate ways for distributing data on ipfs, imo pubsub should only add stuff that you can't do with the current system (realtime notifications)
<zignig>
ion: pin the latest , or validate and republish the validation through ipns.
<multivac>
[WIKIPEDIA] Condorcet method | "A Condorcet method (English pronunciation: /kɒndɔrˈseɪ/) is any election method that elects the candidate that would win by majority rule in all pairings against the other candidates, whenever one of the candidates has that property. A candidate with that property is called a Condorcet winner (named..."
tsenart has quit [Quit: My Mac has gone to sleep. ZZZzzz…]
<Gaboose_>
davidar: fully agree, my point from the start :) the reliability is best leveraged from some side technology
<ion>
zignig: That’s the directory object achin created to link to the archives published by others, isn’t it? He replaced it with a HTML page. Now one can easily pin the index recursively without pinning all the archives. https://ipfs.io/ipns/em32.net/archives/
<erikj`>
Ion: you'd probably get notified for both
<erikj`>
Ah
Stard0g101 has joined #ipfs
* erikj`
should probably attempt to get up to speed before pontificating
voxelot has joined #ipfs
Ceratophrys has quit [Ping timeout: 240 seconds]
<davidar>
Erik: if two messages had the same timestamp, you'd probably do something like ordering them lexigraphically (arbitrary, but deterministic)
gamemanj has joined #ipfs
<davidar>
s/ig/icog
<multivac>
davidar meant to say: Erik: if two messages had the same timestamp, you'd probably do something like ordering them lexicographically (arbitrary, but deterministic)
Stard0g1011 has joined #ipfs
<erikj`>
If you had a sinlge entity keeping track of HEAD, sure
<davidar>
Erik: well, the idea with CRDTs is that everyone will converge onto a single HEAD without a single server tracking it
Stard0g101 has quit [Ping timeout: 240 seconds]
ilyaigpetrov has quit [Remote host closed the connection]
<davidar>
you just have to make sure everyone merges things commutatively (and deterministically)
ilyaigpetrov has joined #ipfs
<davidar>
Erik: does matrix guarantee everyone ends up with the same history, or can there be ordering differences?
* erikj`
will have a read when not walking
tsenart has quit [Quit: My Mac has gone to sleep. ZZZzzz…]
<erikj`>
Everything is a DAG so it only has partial ordering, but that is consistent
<erikj`>
In matrix
<ilyaigpetrov>
I understand you guys are hot about pubsub, but could you please tell me: in your pubsub to send message tagged {#ipfs #matrix} ( http://git.io/vCGsP ) you have to send to #ipfs, #matrix and to #ipfs-matrix?
<Gaboose_>
ilyaigpetrov: yes, or more specifically setT(#ipfs #matrix), setR(#ipfs #matrix) -> setT(#ipfs #matrix #ipfs-matrix) and in another case setT(), setR(#ipfs #matrix) -> setT(#ipfs-matrix)
gaboose has joined #ipfs
Stard0g1011 has quit [Quit: Leaving.]
<davidar>
Erik: looking at the animation on the homepage, matrix already seems to be following a CRDTish model anyway
domanic has joined #ipfs
<ilyaigpetrov>
So has the idea any value for your pubsub? You see it as just another parlance for message routing you won't use?
<ilyaigpetrov>
I know I can't contribute to your go code, but this idea is something I care about to contribute
Albert has joined #ipfs
therealplato1 has quit [Quit: Leaving.]
<davidar>
ilyaigpetrov (IRC): yes, people have already answered that question several times
<davidar>
tagging doesn't need to be integrated into the pubsub system, it can be easily implemented on top as gaboose_ suggests
<ilyaigpetrov>
davidar: thanks, then I was blind in the glow of praising myself :-p
<ilyaigpetrov>
I was messing with this idea day or two thinking it's something special I came up with
jager_ has quit [Quit: No Ping reply in 180 seconds.]
jager has joined #ipfs
therealplato has joined #ipfs
therealplato has quit [Ping timeout: 244 seconds]
pfraze has joined #ipfs
Spinnaker has quit [Ping timeout: 256 seconds]
Spinnaker has joined #ipfs
zz_r04r is now known as r04r
domanic has quit [Ping timeout: 240 seconds]
voxelot has quit [Ping timeout: 265 seconds]
hjoest has quit [Ping timeout: 264 seconds]
pfraze has quit [Remote host closed the connection]
domanic has joined #ipfs
hjoest has joined #ipfs
Encrypt has joined #ipfs
Algolkm has quit [Ping timeout: 268 seconds]
tsenart has joined #ipfs
tsenart has quit [Quit: My Mac has gone to sleep. ZZZzzz…]
domanic has quit [Ping timeout: 264 seconds]
tsenart has joined #ipfs
ilyaigpetrov has quit [Remote host closed the connection]
crossdiver has joined #ipfs
<crossdiver>
untz untz untz it's saturday
<achin>
caturday
_whitelogger has quit [Remote host closed the connection]
_whitelogger_ has joined #ipfs
simpbrain has joined #ipfs
hjoest has quit [Ping timeout: 264 seconds]
hjoest has joined #ipfs
hjoest has quit [Ping timeout: 264 seconds]
Encrypt has quit [Quit: Quitte]
fazo has joined #ipfs
hjoest has joined #ipfs
tsenart has quit [Quit: My Mac has gone to sleep. ZZZzzz…]
tsenart has joined #ipfs
pfraze has joined #ipfs
<ipfsbot>
[go-ipfs] CaioAlonso closed pull request #1820: Adds -H as short option for 'ipfs add --hidden' (master...short-hidden-option) http://git.io/vCO7p
<ipfsbot>
[go-ipfs] CaioAlonso opened pull request #1825: Adds -H as short option for 'ipfs add --hidden' (master...short-hidden-opt) http://git.io/vCGMS
<CaioAlonso>
damn you, GitCop
danslo has quit [Quit: Leaving.]
captain_morgan has joined #ipfs
pfraze has quit [Remote host closed the connection]
<whyrusleeping>
bret: noice. I dont know if youve run into it before, but the random crashing bug that people were seeing on raspberry pis has been fixed
<bret>
woot!
<bret>
whyrusleeping: i applied the workaround
<bret>
some thread setting?
<bret>
should I undo that?
<whyrusleeping>
uhm... that actually might be a different issue, lol
<whyrusleeping>
i would probably keep that one on for now
<bret>
whyrusleeping i think it was the IPFS_REUSEPORT=false
<bret>
env var
<bret>
is that still needed?
<whyrusleeping>
oooh, yeah. That hasnt been fixed yet
<bret>
thats what would lead crashing on mine
<bret>
ipfs for days now
<bret>
/weeks/months
<whyrusleeping>
okay, others reported random panics saying 'slice index out of bounds'
<spikebike>
I think the pi is perfect for p2p nodes that require a constant connection and would allow more mobile clients to check in without the cpu/battery/network load
<spikebike>
the pi2 is a hell of a little computer
<ion>
Personal zero-setup IPFS gateways for mobile devices in the form of a wall wart, anyone?
<spikebike>
sure wall wart, router, or something hockey puck sized
lgierth has quit [Ping timeout: 246 seconds]
rossjones has quit [Ping timeout: 246 seconds]
gaboose has quit [Quit: Connection closed for inactivity]
<spikebike>
the router is nice because it's a perfect place for a proxy/cache/gateway
giodamelio has quit [*.net *.split]
a3nm has quit [*.net *.split]
m3s has quit [*.net *.split]
Sargun has quit [*.net *.split]
alive has quit [*.net *.split]
voxelot has quit [Ping timeout: 260 seconds]
a3nm has joined #ipfs
<lgierth_>
spikebike: the raspi and raspi2 are really bad at crypto
lgierth_ is now known as lierth
lierth is now known as lgierth
<lgierth>
they're far behind x86
<spikebike>
really bad how?
<achin>
slow?
<spikebike>
well sure, x86's are way more expensive and generate way more heat
<lgierth>
yep
Sargun_ has quit [Changing host]
Sargun_ has joined #ipfs
Sargun_ is now known as Sargun
<spikebike>
so sure not idea for encode/decode of 4k streams
<spikebike>
plenty to keep up with normal stuff though
<lgierth>
just saying that it's not the perfect node, it's all tradeoffs
<lgierth>
i.e. the cpu will likely be the bottleneck if you hook it up to a good internet connection
<spikebike>
what do you think the pi2 would be too slow for encryption wise?
<lgierth>
i shouldn't have said "really bad" above, hehe
<lgierth>
my experience with these embedded boards comes from cjdns, ipfs might actually be a bit different since it's all in userspace -- cjdns has a tun interface so it needs to copy packets between userspace and kernel
uhhyeahbret has quit [Quit: WeeChat 1.3]
<achin>
i actually think disk will be a bottleneck too, for an rpi2
<lgierth>
long story short, you might be disappointed by the performance of arm boards
patcon has quit [Ping timeout: 260 seconds]
<spikebike>
I've used em a fair bit and have been impressed
<spikebike>
guess it's all in the expectations
<spikebike>
not tinkered with cjdns though
<spikebike>
I put a 64GB samsung microsd card in mine (fairly fast card), seems pretty snappy for ssh, minecraft, even mathematica
<spikebike>
certainly such slow cores make multithreading more important
<lgierth>
don't sdcards have this corruption issue on power loss?
<spikebike>
I heard mostly about that in regards to crappy consumer ssds
<spikebike>
not aware of any issues with quality microsds and linux
<spikebike>
sure you might lose the last few writes, but linux+ext4 or whatever is reasonably robust
<spikebike>
not saying you should run a bank on one
<lgierth>
:)
<lgierth>
on a different note, any idea under which circumstances a golang net/http server would respond with a status code of 0?
<lgierth>
ipfs.io gateway telemetry shows 0 responses every now and then
uhhyeahbret has joined #ipfs
<ion>
For banks, you should choose AS/400 and COBOL.
kytv has joined #ipfs
<spikebike>
lgierth: as a sample I copied 5 100MB files into a single file (1GB of I/O) and it did so at 19MB/sec. Single threaded md5sum of 500MB, 83MB/sec. scp of 500MB file over wifi = 3MB/sec
<spikebike>
I suspect the last is my slow wifi more than a slow pi
patcon has joined #ipfs
<lgierth>
sounds quite ok
<spikebike>
Doing sha256 for 3s on 8192 size blocks: 9558 sha256's in 3.00s
<spikebike>
(single core)
<spikebike>
for a reasonable checksum/encryption load I suspect a Pi2 would basically keep up with most internet connections in the usa
<spikebike>
sure those lucky enough to have gige uplinks might want something better.
<spikebike>
I've got a netgear R7000 that I'm particularly impressed with
<lgierth>
yeah you're probably right
<spikebike>
it's got a 4xgigE switch, 256MB ram, and 2 radios
<spikebike>
46MB/sec for AES-256
<spikebike>
(pi2)
<lgierth>
meh no oss drivers for the r7000's wifi...
hjoest2 has quit [Quit: Yaaic - Yet another Android IRC client - http://www.yaaic.org]
<spikebike>
lgierth: if you find a killer 802.11ac router with >= 256MB memory and open drivers let me know
<spikebike>
Doing 1024 bit public rsa's for 10s: 26168 1024 bit public RSA's in 10.00s
<spikebike>
given how heavy public key crypto is, and tha'tsjust 1/4th ofthe CPUs that's not bad at all
<spikebike>
so definitely use something thread friendly like go ;-)
amade has quit [Quit: leaving]
hjoest has joined #ipfs
mercora has joined #ipfs
<spikebike>
Doing 521 bit sign ecdsa's for 10s: 1330 521 bit ECDSA signs in 10.00s
<spikebike>
Doing 521 bit verify ecdsa's for 10s: 244 521 bit ECDSA verify in 10.03s
<spikebike>
with all that said, I've not done any IPFS benchmarks.
VictorBjelkholm has quit [Ping timeout: 264 seconds]
kpcyrd has quit [Ping timeout: 250 seconds]
jez0990_ has quit [Ping timeout: 256 seconds]
lgierth has quit [Ping timeout: 265 seconds]
acous has quit [Ping timeout: 265 seconds]
kragniz has quit [Ping timeout: 264 seconds]
Algolkm has quit [Ping timeout: 250 seconds]
kpcyrd has joined #ipfs
jez0990 has joined #ipfs
kragniz has joined #ipfs
lgierth has joined #ipfs
acous has joined #ipfs
nomoremoney has joined #ipfs
ekroon_ has joined #ipfs
crossdiver_ has joined #ipfs
mrrrgn_ has joined #ipfs
<mercora>
hey all and thanks to those who made ipfs a thing ^^
<mercora>
i wondered if it is possible to have the daemon run on another machine and use it via the cli to add objects to the network
okket_ has joined #ipfs
_fil__ has joined #ipfs
orzo_ has joined #ipfs
rektide has joined #ipfs
gwollon has joined #ipfs
gwollon has joined #ipfs
gwollon has quit [Changing host]
<mercora>
im somewhat concerned about running the daemon on my laptop as it causes quite some amount of packets and i thing thats kind of an issue wor wireless links
risk_ has joined #ipfs
<lgierth>
mercora: yeah the CLI talks to the api on port 5001
ogd_ has joined #ipfs
<lgierth>
mercora: if you can secure that, it should be fine. add/cat performance might take a hit obviously :)
tibor__ has joined #ipfs
<mercora>
lgierth: currently when i "add" something without the daemon running it seems to store them somewhere offline... how can i configure it so it uses the remote daemon?
besenwesen_ has joined #ipfs
atgnag_ has joined #ipfs
crossdiver has quit [*.net *.split]
besenwesen has quit [*.net *.split]
risk has quit [*.net *.split]
kyledrake has quit [*.net *.split]
atgnag has quit [*.net *.split]
anderspree has quit [*.net *.split]
tibor has quit [*.net *.split]
ekroon has quit [*.net *.split]
zml has quit [*.net *.split]
rabbitface2 has quit [*.net *.split]
_fil_znc has quit [*.net *.split]
gwillen has quit [*.net *.split]
rektide_ has quit [*.net *.split]
okket has quit [*.net *.split]
brixen has quit [*.net *.split]
_fil_ has quit [*.net *.split]
mrrrgn has quit [*.net *.split]
SoreGums has quit [*.net *.split]
ogd has quit [*.net *.split]
orzo has quit [*.net *.split]
_p4bl0 has quit [*.net *.split]
gozala has quit [*.net *.split]
bren2010 has quit [*.net *.split]
juul has quit [*.net *.split]
GonZo2000 has quit [*.net *.split]
mrrrgn_ is now known as mrrrgn
crossdiver_ is now known as crossdiver
GonZo2000 has joined #ipfs
GonZo2000 has joined #ipfs
GonZo2000 has quit [Changing host]
<lgierth>
mercora: yeah, add and a handful other commands don't require a daemon
bren2010 has joined #ipfs
brixen_ has joined #ipfs
risk_ is now known as risk
<lgierth>
mercora: i'm looking for the option to make it talk to a remote api, i think it should be --api=
<mercora>
lgierth: im somewhat confused... in order to "cat" those files from the network the daemon has to be running or not?
ekroon_ is now known as ekroon
<mercora>
lgierth: i will install it somewhere pi and try