<lgierth>
also i'm $watching the weekly repo now... not sure why i didn't automatically
rendar has quit [Quit: std::lower_bound + std::less_equal *works* with a vector without duplicates!]
<codehero>
hm. seems like the ipfs daemon still can't handle disconnects :D
<noffle>
wrote ipfs-hyperlog today (https://github.com/noffle/ipfs-hyperlog). it's a drop-in hyperlog replacement that creates nodes and hash links that are binary-compatible with ipfs. I'm planning to use the hyperlog ecosystem of modules (hyperkv, swarmlog) on top to build out a pubsub protoype for ipfs.
<voxelot>
noffle: thanks for doing that! i've been trying to wrap my head around how pub/sub can work on ipfs, good to have an example to study
<codehero>
ummm
<codehero>
"ipfs mount" just gave me 'Error: could not resolve name.'
<noffle>
voxelot: it's more of a building block. there's a few more pieces that need to happen on top
<noffle>
codehero: this happens sometimes when your ipns record isn't published widely enough
<noffle>
codehero: daemon online? how many peers?
M-matthew has joined #ipfs
M-david has joined #ipfs
M-davidar has joined #ipfs
M-nated has joined #ipfs
<noffle>
codehero: it tries to fetch from 16 peers
M-whyrusleeping has joined #ipfs
M-osener has joined #ipfs
M-jbenet has joined #ipfs
<noffle>
codehero: we're working on this though -- whyrusleeping has some ideas for ipns improvements I hear
M-amstocker has joined #ipfs
M-eternaleye has joined #ipfs
M-trashrabbit has joined #ipfs
<codehero>
demon is online i think. how do i check peers?
<noffle>
codehero: try republishing and wait for a little bit
M-ikreymer has joined #ipfs
M-mistake has joined #ipfs
M-oddvar has joined #ipfs
<noffle>
ipfs swarm peers
<noffle>
ipfs swarm peers | wc -l
<codehero>
43 peers
M-kode54 has joined #ipfs
M-hash has joined #ipfs
bsm117532 has quit [Killed (rajaniemi.freenode.net (Nickname regained by services))]
M-cryptix has joined #ipfs
M-victorm has joined #ipfs
M-edrex has joined #ipfs
M-erikj has joined #ipfs
M-jgrowl has joined #ipfs
M-roblabla has joined #ipfs
<codehero>
how can it not resolve its own id?
bsm1175321 has joined #ipfs
M-harlan has joined #ipfs
M-kegan has joined #ipfs
M-mubot has joined #ipfs
M-davidar-test has joined #ipfs
M-jfred has joined #ipfs
M-rschulman1 has joined #ipfs
M-giodamelio has joined #ipfs
M-alien has joined #ipfs
M-cocci has joined #ipfs
M-zick has joined #ipfs
M-staplemac has joined #ipfs
M-Ingo has joined #ipfs
M-GereG has joined #ipfs
M-fil has joined #ipfs
M-ryanwilliams has joined #ipfs
M-Peer2Peer has joined #ipfs
M-rschulman has joined #ipfs
<codehero>
huh. i can't cat stuff either
M-matthew has quit [Quit: node-irc says goodbye]
M-mistake has quit [Quit: node-irc says goodbye]
<codehero>
now i can mount
<codehero>
i deleted .ipfs and did init
<codehero>
sigh
voxelot has quit [Ping timeout: 240 seconds]
<noffle>
codehero: ipfs mount is really careful and tries to ensure it has the latest ipns entry
<noffle>
i.e. consistency over availability
<noffle>
eventually we'll be able to have more flexibility and let the user choose that trade-off for themselves
<codehero>
ah, okay
<daviddias>
noffle: can you send me an example? I've been testing against go-ipfs serialized merkledag nodes and so far, everything matches (data, serialized format, links, hashes, etc)
<ipfsbot>
[go-ipfs] whyrusleeping created feat/port-zero-sharness (+1 new commit): https://git.io/vgPPJ
<ipfsbot>
go-ipfs/feat/port-zero-sharness 4904242 Jeromy: trying to use port zero for ipfs daemon in sharness tests...
compleatang has quit [Client Quit]
<lgierth>
noffle: i've updated since
<lgierth>
:)
<lgierth>
it has ipfs-update, dist.ipfs.io, and build from source now
<ipfsbot>
[go-ipfs] whyrusleeping opened pull request #2335: trying to use port zero for ipfs daemon in sharness tests (master...feat/port-zero-sharness) https://git.io/vgPPL
r04r is now known as zz_r04r
HostFat has quit [Quit: Leaving]
nekrodesk has joined #ipfs
nekrodesk has quit [Max SendQ exceeded]
nekrodesk has joined #ipfs
chriscool has joined #ipfs
tmg has quit [Quit: leaving]
dignifiedquire has quit [Quit: Connection closed for inactivity]
Soft has quit [Ping timeout: 256 seconds]
Soft has joined #ipfs
<ipfsbot>
[go-ipfs] chriscool pushed 1 new commit to appveyor-fixes: https://git.io/vgPXl
<ipfsbot>
go-ipfs/appveyor-fixes 59aabfe Christian Couder: appveyor: use a POSIX prereq...
Soft has quit [Ping timeout: 245 seconds]
jaboja has quit [Ping timeout: 250 seconds]
cSmith has joined #ipfs
Matoro has quit [Ping timeout: 248 seconds]
Soft has joined #ipfs
ppham has joined #ipfs
kaiza has joined #ipfs
<domanic>
Question: Is there a document that explains what is the nature of the changes between v03 and v04, I saw the upgrade doc, but it doesn't say what changed except line: "lots of great changes and enhancements"
<domanic>
jbenet, ^
Soft has quit [Ping timeout: 264 seconds]
Soft has joined #ipfs
randomguy has quit [Ping timeout: 250 seconds]
Soft has quit [Ping timeout: 240 seconds]
joshbuddy has joined #ipfs
multivac has quit [Ping timeout: 240 seconds]
multivac has joined #ipfs
dylanPowers has quit [Quit: No Ping reply in 180 seconds.]
nekrodesk has quit [Excess Flood]
dPow has joined #ipfs
Soft has joined #ipfs
ppham has quit [Remote host closed the connection]
ppham has joined #ipfs
Oatmeal has quit [Ping timeout: 252 seconds]
Soft has quit [Ping timeout: 264 seconds]
Soft has joined #ipfs
Oatmeal has joined #ipfs
<ipfsbot>
[go-ipfs] chriscool force-pushed appveyor-fixes from 59aabfe to b87fac8: https://git.io/vgiwy
<ipfsbot>
go-ipfs/appveyor-fixes b87fac8 Christian Couder: appveyor: use a POSIX prereq...
pfraze has joined #ipfs
Oatmeal has quit [Ping timeout: 240 seconds]
voxelot has quit [Ping timeout: 250 seconds]
ppham has quit [Remote host closed the connection]
ppham has joined #ipfs
nekrodesk has joined #ipfs
nekrodesk has quit [Max SendQ exceeded]
Oatmeal has joined #ipfs
nekrodesk has joined #ipfs
nekrodesk has quit [Max SendQ exceeded]
nekrodesk has joined #ipfs
nekrodesk has quit [Max SendQ exceeded]
nekrodesk has joined #ipfs
nekrodesk has joined #ipfs
joshbuddy has quit [Ping timeout: 240 seconds]
ulrichard has joined #ipfs
lothar_m has quit [Quit: WeeChat 1.4-dev]
joshbuddy has joined #ipfs
ecloud has quit [Ping timeout: 260 seconds]
ecloud has joined #ipfs
gigq has quit [Ping timeout: 240 seconds]
gigq has joined #ipfs
pfraze has quit [Remote host closed the connection]
<ipfsbot>
[go-ipfs] chriscool pushed 1 new commit to appveyor-fixes: https://git.io/vgPdC
<ipfsbot>
go-ipfs/appveyor-fixes 7d4eb32 Christian Couder: test-lib: add test_path_cmp()...
mildred has joined #ipfs
* jbenet
domanic: we'll be writing a blog post about this
<Qwertie>
jbenet, You said a major browser is implementing ipfs. Can you say which one?
mildred1 has joined #ipfs
<jbenet>
Qwertie: No, sorry. I appreciate the enthusiasm, but please stop asking about it
<Qwertie>
I missed your reply last time. sorry
<jbenet>
richardlitt: we should ask vijayee about moving that repo to the org-- (or we can fork it and stay up to date with its changes). i'd like to keep all the huge officially-ipfs things inside the org for easy backing up + finding + licensing + project-repos listing + code of conduct coverage + ... etc.
<jbenet>
Qwertie: no problem :) -- it's more a general statement. i get asked about it often but cant say anything.
Soft has quit [Ping timeout: 240 seconds]
Soft has joined #ipfs
Soft has quit [Ping timeout: 240 seconds]
M-jon has quit [Quit: node-irc says goodbye]
Soft has joined #ipfs
M-eternaleye is now known as eternaleye
eternaleye has joined #ipfs
eternaleye has quit [Changing host]
Soft has quit [Ping timeout: 264 seconds]
IlanGodik has joined #ipfs
e-lima has joined #ipfs
Soft has joined #ipfs
ylp1 has joined #ipfs
erl has joined #ipfs
Soft has quit [Ping timeout: 245 seconds]
Soft has joined #ipfs
rendar has joined #ipfs
gigq has quit [Ping timeout: 264 seconds]
gigq has joined #ipfs
srenatus has joined #ipfs
nekrodesk has quit [Max SendQ exceeded]
hashcore has joined #ipfs
mildred has quit [Quit: Leaving.]
dignifiedquire has joined #ipfs
erl has quit [Quit: Leaving]
mildred has joined #ipfs
Soft has quit [Ping timeout: 264 seconds]
Soft has joined #ipfs
disgusting_wall has quit [Quit: Connection closed for inactivity]
hashcore has quit [Ping timeout: 240 seconds]
<ipfsbot>
[go-ipfs] chriscool pushed 1 new commit to appveyor-fixes: https://git.io/vgXe4
<ipfsbot>
go-ipfs/appveyor-fixes a7c683f Christian Couder: t0020: improve error message...
Soft has quit [Ping timeout: 248 seconds]
Soft has joined #ipfs
joshbuddy has quit [Quit: joshbuddy]
Tv` has quit [Quit: Connection closed for inactivity]
joshbuddy has joined #ipfs
ppham has quit [Remote host closed the connection]
ppham has joined #ipfs
joshbuddy has quit [Ping timeout: 248 seconds]
joshbuddy has joined #ipfs
Soft has quit [Ping timeout: 264 seconds]
<The_8472>
can a node embed signatures into ipfs-paths to preovide reverse mappings to ipns? i.e. to say "this data is or may have been available under /ipns/..."?
<jbenet>
The_8472 yeah-- well the goal is to do a bunch of object signing stuff, where objects themselves can be signed (by a "signature object" that links to it).
<The_8472>
i guess for unixfs that could be put into metadata?
<jbenet>
daviddias: good to hear-- since this has caused so much pain to move to that :)
<dignifiedquire>
jbenet: are you still awake Oo
martinBrown has joined #ipfs
hashcore has quit [Ping timeout: 240 seconds]
vaelys_ is now known as vaelys
Soft has quit [Ping timeout: 264 seconds]
Soft has joined #ipfs
randomguy has quit [Quit: Leaving.]
Soft has quit [Read error: Connection reset by peer]
amiller has quit [Ping timeout: 256 seconds]
domanic has quit [Ping timeout: 240 seconds]
mvollrath has quit [Max SendQ exceeded]
amiller has joined #ipfs
amiller is now known as Guest58194
martinkl_ has joined #ipfs
mvollrath has joined #ipfs
<jbenet>
And the IPLD spec PR lands!! https://github.com/ipfs/specs/pull/37 thank you very much to everyone, in particular to mildred who powered through all of the last huge problems and made it happen
<Kubuxu>
lgierth: could you run the same nginx proxies as you run on :80 but also on :8080 or some other port that we all agree on and design is as to be new standard local gateway port??
drnet has joined #ipfs
Encrypt has quit [Quit: Quitte]
hashcore has quit [Ping timeout: 252 seconds]
pfraze has joined #ipfs
<Kubuxu>
lgierth: or not scrap that
<lgierth>
yeah i don't understand :)
drnet has quit [Quit: Leaving]
<Kubuxu>
It is a shame that IPv6 doesn't have bigger localhost subnet.
<Kubuxu>
But IPv4 has
<Kubuxu>
I might have found solution for all our problems, no need for gateway redirects and so on.
<Kubuxu>
Bind local gateway to 127.0.0.x and add ipfs.io 127.0.0.x, gateway.ipfs.io 127.0.0.x to /etc/hosts
<Kubuxu>
lgierth: ^^
<Kubuxu>
Then you can just bind it on port 80
zorglub27 has quit [Ping timeout: 264 seconds]
<Kubuxu>
127.0.0.1/8 is whole loopback
<Kubuxu>
So we can just choose random address form it and bind.
<Kubuxu>
Problem is 80 < 1024 so you need root to bind it on linux but IIRC there is some workaround for it.
<The_8472>
there's a sysctl that controls the port range, but changing that requires root
<Kubuxu>
Yes, but this is one time change/config. Point is that IPFS doesn't need to start as root.
<The_8472>
eh, start as root, drop privs is probably saner than changing system-wide configuration
<Kubuxu>
but running local gateway on port 80 would solve so many problems.
<The_8472>
such as?
<The_8472>
also, you still wouldn't have ssl support
<Kubuxu>
Ok then it might not have been so good idea after all.
jaboja has joined #ipfs
<lgierth>
Kubuxu: that won't work
<lgierth>
we'll have HSTS enabled soon so browser will force https://ipfs.io
<lgierth>
and gateway.ipfs.io will just be a redirect to ipfs.io
<Kubuxu>
I haven't taken https into equation at all.
hashcore has joined #ipfs
zz_r04r is now known as r04r
m0ns00n has joined #ipfs
<lgierth>
there needs to be another way to find a local gateway
<lgierth>
are there any mdns apis in the browsers?
<lgierth>
how do you detect a local gateway right now, polling :8080/api/v0/version ?
<The_8472>
udp sockets, use multicast
Encrypt has joined #ipfs
ppham has quit [Remote host closed the connection]
<tmg>
this first gx deps build is painful
ppham has joined #ipfs
hashcore has quit [Ping timeout: 252 seconds]
m0ns00n has quit [Quit: undefined]
hashcore has joined #ipfs
hashcore has quit [Ping timeout: 250 seconds]
<daviddias>
lgierth there is a W3C draft for mdns in the browser, but I don't know of any implementation or anyone using it
<The_8472>
why so overcomplicated anyway? pick a multicast group, a port and glue together a packet format with some duct tape and you have your custom discovery protocol
jaboja has quit [Ping timeout: 272 seconds]
<Kubuxu>
The_8472: you have to ask ICANN to assign you multicast group
Qwertie has quit [Ping timeout: 260 seconds]
<The_8472>
there are private-use blocks
<The_8472>
or SSDP could be used, it has an assigned address
<Kubuxu>
or mdns could be used
<The_8472>
more complicated to implement than ssdp
<The_8472>
basically, why mdns?
<Kubuxu>
why not, it works, you have DNS framework already there
<The_8472>
xml works too
<The_8472>
doesn't mean you use xml for everything
zorglub27 has joined #ipfs
<lgierth>
mdns is already implemented in go-ipfs
<lgierth>
i have nothing against udp multicast
<lgierth>
(actually like it very much)
<lgierth>
where it gets really interesting is ipv6 link-local multicast
<lgierth>
but i doubt one can do that from a browser addon, let alone a website
<lgierth>
(you need ifnames of network interfaces)
<lgierth>
i.e. ping6 ff02::1%eth0
jaboja has joined #ipfs
<lgierth>
browser at least don't support ipv6 link-local addresses in the address bar
ppham has quit [Remote host closed the connection]
ppham has joined #ipfs
ppham has quit [Remote host closed the connection]
e-lima has quit [Ping timeout: 264 seconds]
ppham has joined #ipfs
Encrypt has quit [Quit: Quitte]
thengeek has joined #ipfs
ppham has quit [Remote host closed the connection]
ppham has joined #ipfs
<thengeek>
hello everyone :)
thengeek has quit [Client Quit]
ppham has quit [Remote host closed the connection]
<Kubuxu>
under a minute...
e-lima has joined #ipfs
<ipfsbot>
[js-ipfs-api] dignifiedquire created greenkeeper-babel-core-6.5.2 (+1 new commit): https://git.io/vg1Bu
<ipfsbot>
js-ipfs-api/greenkeeper-babel-core-6.5.2 2bc51f1 greenkeeperio-bot: chore(package): update babel-core to version 6.5.2...
<ipfsbot>
[js-ipfs-api] dignifiedquire deleted greenkeeper-babel-core-6.5.2 at 2bc51f1: https://git.io/vg1Rd
mirek has left #ipfs [#ipfs]
<ipfsbot>
[js-ipfs-api] dignifiedquire created greenkeeper-babel-plugin-transform-runtime-6.5.2 (+1 new commit): https://git.io/vg1E8
<ipfsbot>
js-ipfs-api/greenkeeper-babel-plugin-transform-runtime-6.5.2 db7b03d greenkeeperio-bot: chore(package): update babel-plugin-transform-runtime to version 6.5.2...
<ipfsbot>
[js-ipfs-api] dignifiedquire deleted greenkeeper-babel-plugin-transform-runtime-6.5.2 at db7b03d: https://git.io/vg1zY
ppham has joined #ipfs
chriscool has joined #ipfs
ppham has quit [Remote host closed the connection]
jaboja has joined #ipfs
ppham has joined #ipfs
ppham has quit [Remote host closed the connection]
ppham has joined #ipfs
<daviddias>
voxelot: are you around?
s_kunk has quit [Ping timeout: 240 seconds]
<voxelot>
daviddias: o/
<voxelot>
getting sidetracked looking at scuttlebutt instead of writing tests :D
<daviddias>
at a real scuttlebutt or the scuttlebutt scuttlebutt? :P
<noffle>
scuttlebutt \o/
<daviddias>
I'm imagining you looking at the bubbles in the Scuttlebutt
<daviddias>
voxelot: I've one another thing that I would love to have with and it should be pretty fun
<voxelot>
haha just checking out noffle's hyperlog version with merkle dags
<daviddias>
to have *help* with
<voxelot>
sure whats up
<daviddias>
I've finishes the file and dir importing using unixfs
<daviddias>
s/finishes/finished
<daviddias>
and now, since we are doing it in JS and we love streams for everything + make it work in the browser
<daviddias>
I want to have a bufferImporter and a streamImporter
<daviddias>
it would be great if you could PR those, because that would mean that the Importing flow looks good
chriscool has quit [Quit: Leaving.]
kaiza has quit [Quit: Leaving]
<daviddias>
data importing tests are kind of very througout, so you should not find nasty bugs with the unixfs impl or the merkledag impl
<daviddias>
Also, if I understand correctly, ipfs-dag is the DAGimpl
<daviddias>
It would be great to have these ideas for cleaner API on core API spec
<daviddias>
s/DAGimpl/DAGNode impl
<noffle>
daviddias: yeah, I saw it was a pretty direct port
<noffle>
but the go impl could use a lot of this kind of love too
<daviddias>
I agree, we can definitely do better
<daviddias>
Ultimately, we would have IPFS implement ions across languages that are modular used with interfaces that match, so that tests can be developed with certain expectations for all of them
<noffle>
do we have a core api spec for this?
<noffle>
yes that'd be rad
zorglub27 has quit [Ping timeout: 264 seconds]
<daviddias>
noffle, we don't have core API spec'ed out, but seems like a good time to start :)
ppham has quit []
<ipfsbot>
[go-ipfs] whyrusleeping pushed 1 new commit to feat/port-zero-sharness: https://git.io/vg1hO
<ipfsbot>
go-ipfs/feat/port-zero-sharness e28d619 Jeromy: clean up test t0062...
<whyrusleeping>
if i can figure this sharness stuff out, it should cut out most of our random test failures
<noffle>
whyrusleeping: context?
arpu has quit [Quit: Ex-Chat]
<whyrusleeping>
the commit i pushed before commenting
<whyrusleeping>
we get a lot of sharness tests that fail because we pick a port thats already in use for the daemon to run on
<lgierth>
it looks like i still need to have go-ipfs in GOPATH for make install
palkeo has joined #ipfs
cemerick has joined #ipfs
<Kubuxu>
probably due to gx using global deps
<Kubuxu>
but I might be wrong
jcd has joined #ipfs
<lgierth>
ok meh, now i linked it into gopath and it's complaining about "no install location [..] outside gopath"
* lgierth
wondering why it go builds and then go installs
<ipfsbot>
[go-ipfs] whyrusleeping pushed 1 new commit to feat/port-zero-sharness: https://git.io/vgMk4
<ipfsbot>
go-ipfs/feat/port-zero-sharness 2669a56 Jeromy: fix gateway tests to work with new port0 fun...
<whyrusleeping>
go-ipfs needs to be in the GOPATH because of the relative import path crap
<whyrusleeping>
which i'm hoping we can get away from with gx...
<whyrusleeping>
eventually, i'd like go-ipfs to just be the cli binary code
<whyrusleeping>
and everything else to be pieced out into separate repos
<whyrusleeping>
now that we're using gx, we can start doing that
<lgierth>
so my setup might be a bit weird. i'm working outside of gopath, but gopath/src/github.com/ipfs/go-ipfs is a symlink to my go-ipfs working dir
<whyrusleeping>
huh
<whyrusleeping>
i'm honestly not sure how that would work out
<whyrusleeping>
i know the other way around works
<lgierth>
it usually does work
<Kubuxu>
It works ok with native go
<lgierth>
i'm on my laptop right now and on the desktop it did work
<lgierth>
without the symlink it of course complained that it couldn't find the imports
<lgierth>
maybe need to set GOBIN?
<lgierth>
i think the install target shouldn't depend on the build target
<lgierth>
install compiles too so it's fine
<lgierth>
yep ok GOBIN=$GOPATH/bin go install works
<Kubuxu>
hmm, I though it would just default to that.
voxelot has quit [Read error: Connection reset by peer]
ehmry has joined #ipfs
ehmry has left #ipfs [#ipfs]
voxelot has joined #ipfs
<lgierth>
ok i have a few amendmends to the make build/install targets, which might or might not be good
<lgierth>
didn't touch the gx target
rendar has quit [Ping timeout: 276 seconds]
<lgierth>
hey dan
<lgierth>
i built an openwrt image with cjdns the other day
<lgierth>
on the other computer though, i'll be home in a few hours
rendar has joined #ipfs
<lgierth>
ups wrong channel
<noffle>
lgierth: does the public gateway hang forever when it can't find a hash? I killed my curl req after 5 mins
<lgierth>
yeah
<lgierth>
the nginx is timeout is large
<lgierth>
we can't really determine that a hash does not exist
<noffle>
yes
<noffle>
it's rough on automated tooling though -- they'd wait forever
<jbenet>
we may want to add smaller default timeouts and only have huge timeouts with a query option or a header
<noffle>
agreed
<jbenet>
(i know im the one pushing for huge timeouts for people in countries with horrible latencies)
<lgierth>
well
<lgierth>
if the tool is gonna pass a timeout query param or header
<lgierth>
it can just go kill the connection on its own
<noffle>
lgierth: timing out after > 5 mins seems like a bad default though
<noffle>
not sure if that's the norm
<noffle>
or the exceptional case
<noffle>
in terms of what people/tools desire
<lgierth>
i don't have strong feelings one way or the other
<lgierth>
just saying that if the client needs to be aware of the never-timeouts anyhow [...]
<lgierth>
having a hash and requesting that from the gateway means the respective object *does* exist
<noffle>
there's no really nice solution; http wasn't designed with "I might have what you're requesting in a while" in mind
<lgierth>
yep :)
<noffle>
lgierth: right, but NOT that it's on the network
<jbenet>
and http today leaves a ton of people unable to use websites with horrible liveness timeouts. (sometimes tls handshakes cannot complete)
<jbenet>
(most of the world has terrible latencies to the backbone)
HostFat has quit [Read error: Connection reset by peer]
<lgierth>
when i'm behind triple-nat it sometimes takes minutes to resolve stuff i add -- refreshing every once in a while works. i dunno i have no strong feelings. agree with what jbenet says about slow connections
<lgierth>
i think we use nginx' defaults for the tls handshake btw
rhalff has joined #ipfs
<noffle>
lgierth: that makes sense
<noffle>
it's a trade off either way, but at least with a long timeout the client can kill manually (basically what you were saying, lgierth)
<noffle>
with a short timeout you cannot win if you have a high lat conn
<noffle>
I wonder if we could have exponential backoff for timeouts? first req timeouts out fast, next one lasts longer, etc. so one-off reqs can fail fast, but users with bad latency can retry and have a chance at winning
<noffle>
though introducing state is often a bad idea
<The_8472>
adaptive timeouts on the client
<The_8472>
just measure the histogram for response times, kill when you have X% confidence that it should by done by that time
<Codebird>
A short timeout may result in an error to the user on the first attempt - but so long as the resolve attempt continues, they can always retry later.
<noffle>
The_8472: not all clients support this; we can't expect them to :(
<The_8472>
noffle, well, if clients don't even support timeouts they obviously don't care about speeds :P
<ipfsbot>
[js-ipfs-api] greenkeeperio-bot opened pull request #210: eslint@2.0.0 breaks build
srenatus has quit [Quit: Connection closed for inactivity]
lothar_m has joined #ipfs
<noffle>
The_8472: it's easy to say that, but even if the software doesn't the end-users do
<noffle>
it still reflects poorly on us
<noffle>
so we need to be defensive
<noffle>
lgierth: did you modify the public gateway? 'go get' is working now!!! :o
<noffle>
go get -v ipfs.io/ipfs/QmdQbpEKwuiZj796Per1eD4AJmXRPSoWH5CMHL64HkREw1/go-multihash.git
* lgierth
whistles innocently
<lgierth>
i did literally nothing
<noffle>
very strange. my local gateway still hangs
<The_8472>
noffle, i wasn't quite serious about it.
<jbenet>
noffle: yes agreed on it being up to us
<noffle>
The_8472: sorry, poe's law :)
<whyrusleeping>
i love waiting around for travis to finish...
<whyrusleeping>
pushed a commit, went to lunch, came back, still waiting for the tests to even start
<jbenet>
i've just been noticing the gateway taking longer to do everything, it coincided roughly with multireq deployment, not sure if its something going wrong there, or just the network dealing with more demand, or the machines being more strained (with 2 nodes)
<noffle>
whyrusleeping: followed by 'retry the tests that are timing out until they go green'
<whyrusleeping>
noffle: lol, hopefully that won't be needed anymore
<jbenet>
whyrusleeping: it will get better with broken out repos. but agreed. lgierth and i can go talk to them when im in berlin and make a strong case for letting us pay for it.
<jbenet>
whyrusleeping: nice! great to hear!
<noffle>
whyrusleeping: really excited about that
<whyrusleeping>
yeeeep!
<whyrusleeping>
so many of our random failures were 'we picked a bad random port'
mkelly has quit [Quit: My Mac has gone to sleep. ZZZzzz…]
<noffle>
whyrusleeping: does that explain timeouts? wouldn't the os let us know immediately?
<whyrusleeping>
well, daemon would fail to startup
<whyrusleeping>
our checks would pass because
<whyrusleeping>
'theres a daemon at that address'
<whyrusleeping>
then that other daemon would die, because its leftover from the previous tests
<whyrusleeping>
then our tests would sit and poll for some endpoint
<jbenet>
so it was that after all-- sigh.
<noffle>
ha
<whyrusleeping>
yeah, super fun
<whyrusleeping>
and the reason it had been getting worse lately is simply because we were running more and more tests
<whyrusleeping>
more tests == more chances for random failures
cemerick has quit [Ping timeout: 272 seconds]
<The_8472>
lgierth, how about changing the pref-utils test module to backup the prefs once at the start and then resetting them before each test?
Encrypt has joined #ipfs
ugjka has joined #ipfs
<voxelot>
dignifiedquire: can i get your input on a weird webpack/node-forge/karma error i'm still having
<whyrusleeping>
'chaosmonkey' isnt trademarked or anything, is it?
<whyrusleeping>
i was thinking of having an 'ipfs-chaosmonkey' that did all sorts of annoying things
<whyrusleeping>
like connect to nodes and leave the sockets half open
<whyrusleeping>
or get partly through a handshake and quit
<noffle>
whyrusleeping: that'd be great
<noffle>
whyrusleeping: we could have *-chaos-monkeys that run amok at all levels of the stack (tcp, utp, http, ipfs)
<whyrusleeping>
yeah, that would be sweet
<noffle>
products just choose a suite of monkeys to clobber their systems with
<whyrusleeping>
dht, bitswap
<The_8472>
i think the term chaosmonkey is usually used for infrastructure/redundancy testing
<whyrusleeping>
The_8472: thats the idea
<The_8472>
otherwise it's more fuzzing, no?
<whyrusleeping>
The_8472: yeah, i guess it could be considered fuzzing
<noffle>
the name probably doesn't matter too much in this case (I really like the term chaosmonkey)
<noffle>
so I'm biased
jaboja has quit [Ping timeout: 250 seconds]
<The_8472>
aiui the monkies at netflix&co attack at a different level. randomly killing processes or VMs
<lgierth>
The_8472: not sure what you mean about pref-utils
<The_8472>
the thing that backups and restores prefs around the tests
<multivac>
[WIKIPEDIA] Cosmic ray | "Cosmic rays are immensely high-energy radiation, mainly originating outside the Solar System. They may produce showers of secondary particles that penetrate and impact the Earth's atmosphere and sometimes even reach the surface. Composed primarily of high-energy protons and atomic nuclei, they are of..."
<The_8472>
don't radioactive lead isotopes in solder cause more bit flips in ram than cosmic ray impacts?
<whyrusleeping>
The_8472: on earth
<The_8472>
hah
<whyrusleeping>
;)
<The_8472>
it's just interplanetary though, not interstellar!
<whyrusleeping>
'things we don't have to worry about on earth but are super shitty in space'
kaiza has joined #ipfs
ashark has quit [Ping timeout: 276 seconds]
<The_8472>
the heliosphere already shields the solar system from cosmic rays to some extent
<The_8472>
but you get nice solar wind instead
<jbenet>
whyrusleeping: yesssss cosmic rays!
<jbenet>
whyrusleeping: it's a thing in papers to always be like "well this is a problem, but the probability is lower than cosmic rays flipping random bits in memory, so it's not an issue".
<The_8472>
i think zfs people would club you to death if you said that
<jbenet>
whyrusleeping: +1 on these tools. be careful before deploying them on the live network, we should first run them in isolation till we're sure it passes. we do have users that depend on this stuff. we need to get above the bar of passing those attacks before running on the live network. but im all for randomly attacking ourselves to ensure our protocols are
<jbenet>
good.
mkelly has joined #ipfs
<whyrusleeping>
>.>
* whyrusleeping
quickly kills a few scripts
<dignifiedquire>
voxelot: what's up?
<lgierth>
and we probably want latency based dns first. currently if a host is down to whatever degree, it'll still get requests
hannes3 has left #ipfs ["Leaving"]
<voxelot>
dignifiedquire: actually might be better after i PR so you can actually see what's happening
<voxelot>
for some reason when i test peer-id through js-ipfs webpack exposes forge.forge().someFunction() and when i test peer-id in the same way alone it exposes the normal forge.someFunction() on the browser
mkelly has quit [Quit: My Mac has gone to sleep. ZZZzzz…]
Qwertie has joined #ipfs
pfraze has quit [Remote host closed the connection]
<ipfsbot>
[go-ipfs] chriscool created improve-windows-doc (+1 new commit): https://git.io/vgM8l
<ipfsbot>
go-ipfs/improve-windows-doc 0e4c47e Christian Couder: This add a lot of steps that were missing and makes it possible to build on Windows....
HostFat has joined #ipfs
ugjka has quit [Remote host closed the connection]
jaboja has joined #ipfs
vijayee_ has joined #ipfs
<ipfsbot>
[go-ipfs] chriscool force-pushed improve-windows-doc from 0e4c47e to 912623e: https://git.io/vgMBJ
<ipfsbot>
go-ipfs/improve-windows-doc 912623e Christian Couder: docs: improve windows.md...
<Kubuxu>
Only one file in cmd/ipfs has to be poisoned like htat
<Kubuxu>
s/htat/that/
disgusting_wall has quit [Quit: Connection closed for inactivity]
<noffle>
might be a good fallback if we can't figure out the compatibility route
f[x] has joined #ipfs
screensaver has joined #ipfs
<ipfsbot>
[js-ipfs-api] dignifiedquire created greenkeeper-karma-0.13.21 (+1 new commit): https://git.io/vgMoY
<ipfsbot>
js-ipfs-api/greenkeeper-karma-0.13.21 e989675 greenkeeperio-bot: chore(package): update karma to version 0.13.21...
joshbuddy has joined #ipfs
palkeo has quit [Ping timeout: 240 seconds]
<Kubuxu>
I think even if we figure out compatibility route we should find transition strategy, as if we make it compatible there won't be incentive to push for local node usage
palkeo has joined #ipfs
Oatmeal has joined #ipfs
jhulten has joined #ipfs
wlan2 has joined #ipfs
<lgierth>
oh hey noffle
<lgierth>
i was under the impression somehow that my pr had merged haha
* wlan2
was trying to follow the example here: /ipfs/QmTkzDwWqPbnAh5YiV5VwcTLnGdwSNsNTn2aDxdXBFca7D/example#/ipfs/QmQwAP9vFjbCtKvD8RkJdCvPHqLQjZfW7Mqbbqx18zd8j7/git/readme.md
<Qwertie>
I tried updating go-ipfs with go get but im getting
<Qwertie>
error: some local refs could not be updated; try running
<Qwertie>
'git remote prune origin' to remove any old, conflicting branches
pfraze has quit [Remote host closed the connection]
<wlan2>
But it does not clone that
<noffle>
lgierth: for multireq?
<noffle>
Kubuxu: I agree: be compatible, but offer a clear upgrade path with obvious benefits
<lgierth>
noffle: yeah... i have a 2xx/3xx vs 4xx/5xx block there
<lgierth>
mmh having second thoughts
<lgierth>
there's this thing where in go-ipfs the request context times out
<lgierth>
and i think that leads to a 404 too
<lgierth>
which it shouldn't of course
<Kubuxu>
but timeout is so long that it shouldn't matter IMO
rod_ has joined #ipfs
<lgierth>
no, different thing
ashark has quit [Ping timeout: 252 seconds]
<lgierth>
i haven't seen a context timeout in a long time though
<lgierth>
whyrusleeping: ^
<Kubuxu>
404 is valid when you access non existing file/dir inside a DAG node.
<rod_>
hello, i just found out about ipfs, wonder if it works from behind nat uplink, can sombody please xplain ?
<Kubuxu>
ipfs is able to penetrate NAT
<Kubuxu>
in most cases
<Kubuxu>
rod_: ^^
<wlan2>
It's slower if you don't have ports exposed, though, rod_
<wlan2>
Either that of my connection is just a fine load of crap. But i'd guess it's a combination of both.
<Kubuxu>
It shouldn't be
<voxelot>
dignifiedquire: if you get the time can you compare js-ipfs and js-peer-id on the issue mentioned before, incoming links
<Kubuxu>
It uses UPnP so it shouldn't affect the performance in any way.
<rod_>
Kubuxu, i have a router with a dynamic ip assigned from isp and behind router i am using 192.168.1.255 net wonder how poeple can acess files from my laptop for example .... do i have to install port forwarding
<wlan2>
rod_: if you use ipfs for it, there is no big need to forward the ports.
<Kubuxu>
rod_: if your router support UPnP IPFS will find its way though.
<rod_>
Kubuxu, ok, thanks then i will give it a try
<Kubuxu>
Just add something on your computer and use https://ipfs.io/ipfs/[HASH] to access it via public gateway
<wlan2>
Kubuxu: Do you know if there might be anything wrong with the "git, even more decentralised" example?
<Kubuxu>
were you adding those files by your self?
<Kubuxu>
nobody might be hosting this hash anylonger
<Kubuxu>
yup it looks like nobody is hosting it anylonger
<wlan2>
No, I was trying my own hash
<Kubuxu>
hmm, then IDK it should work
<wlan2>
Git complains about no link named "refs", but I think it might be a problem of ipfs not adding some of the files?
domanic has quit [Ping timeout: 276 seconds]
<wlan2>
Trying to do from a local repo fom scratch, not a hosted one.
<Kubuxu>
Hmm, it might be fact that IPFS is now not adding hidden files
<Kubuxu>
try `ipfs add -rH .
<Kubuxu>
`
<Kubuxu>
wlan2: ^^^
<wlan2>
Ok, trying again.
<lgierth>
Kubuxu: yeah exactly, 404 is totally legitimiate when that link doesn't exist. on the other hand, i'm pretty sure these context timeouts also result in a 404 and that's not a legitimate response
<rod_>
Kubuxu, get an error when trying to ipfs mount ....
<ipfsbot>
[js-ipfs-api] dignifiedquire deleted greenkeeper-karma-0.13.21 at e989675: https://git.io/vgM9R
<Kubuxu>
rod_: I am off too sleep, check if fuse kernel module is loaded, you have to own the mount-point, your user has to have read permissions to /etc/fuse.conf and few more requirements
<wlan2>
Kubuxu: Beautiful! Possibly the best license for less than 300 lines!
<ipfsbot>
[go-ipfs] whyrusleeping pushed 1 new commit to master: https://git.io/vgM9D
<voxelot>
test times out on creating RSA key pair though
<daviddias>
woooooo :)
<daviddias>
strange
<daviddias>
takes more than 2 seconds?
<voxelot>
yeah
<daviddias>
how much does it take?
<voxelot>
sometimes, depends
<voxelot>
usually between 203 seconds
<voxelot>
2-3
<wlan2>
This may seem dumb to ask from someone able to use git, but, how is the js implementation going?
<voxelot>
i also put it into js-ipfs, but waiting to PR that until the peer-id merges so my github doesn't have to be in the package? https://github.com/nginnever/js-ipfs
<palkeo>
hey
<palkeo>
I just made a quick hack to see the blocks passing by your node