<zignig>
whyrusleeping: knows how it works, it's black magic as far as i'm concerned ;)
chriscool has quit [Ping timeout: 256 seconds]
<kyledrake>
Oh thank god, I've finally found a project I like that uses protobufs
<kyledrake>
thx zignig ;)
m0ns00n has joined #ipfs
mildred has quit [Ping timeout: 272 seconds]
mildred has joined #ipfs
grncdr has quit [Ping timeout: 276 seconds]
<zignig>
np !
mitchty has joined #ipfs
zabirauf_ has joined #ipfs
domanic has joined #ipfs
zabirauf_ has quit [Ping timeout: 264 seconds]
anshukla has joined #ipfs
zootella has joined #ipfs
elima_ has quit [Remote host closed the connection]
elima has joined #ipfs
anshukla has quit [Remote host closed the connection]
zabirauf_ has joined #ipfs
Wallacoloo has joined #ipfs
<cryptix>
hello :)
<jbenet>
hello o/
<kyledrake>
jbenet I'm ipfs block get DIR_OBJ to get the raw directory data. Does that seem correct to you? The protobuf libraries are not happy with it.
<kyledrake>
s/dir_obj/hash
<kyledrake>
"Wrong type `2` given for (Type). Expected 0" <- gotten this on 2 protobuf parsers now.
<jbenet>
kyledrake: need to pass the encoding flag but yeah
<jbenet>
ipfs object get --enc protobuf $root -- I think
<jbenet>
Can try: ipfs blocks get $root
<ipfsbot>
[go-ipfs] cryptix force-pushed feat/httpApiTcpPort0 from cea5aa2 to 4b337bb: http://git.io/vkWm3
<ipfsbot>
go-ipfs/feat/httpApiTcpPort0 07b3415 Henry: http endpoints: dont print before listen...
<cryptix>
jbenet: #1287 should be fine now (i hope...)
<jbenet>
kyledrake: that's embedded in the data segment of a merkledag.proto
<kyledrake>
Okay, that worked. Thanks!
* kyledrake
has never actually used protobufs before
<kyledrake>
I'm trying your suggestion of direct node manipulation for file adds. This is really clean, and I think I can figure this out.
<kyledrake>
This should make it efficient enough that I can use the HTTP API directly
inconshr_ has joined #ipfs
inconshreveable has quit [Ping timeout: 265 seconds]
<jbenet>
oh kyledrake check out `ipfs object patch` it landed this week
<jbenet>
kyledrake we don't yet support "union dirs" (a dir sharded into multiple objects) so your root object is going to be big (unless you shard it yourself by adding some tiering (like "foobar1023 -> "fo/ob/ar1023")
<kyledrake>
jbenet I'll look at those, thanks. I think for the per-site partition the size will be reasonable. Not so much for the entire thing though.
<kyledrake>
jbenet certainly faster than running ipfs add -r every time a single file change is made
<jbenet>
kyledrake yeah if you already have the knowledge of which files changed helps too.
ei-slackbot-ipfs has quit [Remote host closed the connection]
hpk has quit [Ping timeout: 272 seconds]
ei-slackbot-ipfs has joined #ipfs
<daviddias>
Helloo! :)
<jbenet>
daviddias o/ :)
<jbenet>
morning!
<cryptix>
jbenet: didnt see object patch - can it add/overwrite objects to a node and give you the new root? i think i could use it then to fix the http writeable api finally
<daviddias>
Morning :)
<cryptix>
hey daviddias :)
<daviddias>
How is everyone today?
* cryptix
loadavg > 9000
<cryptix>
but fine actually
<ipfsbot>
[go-ipfs] jbenet created gpe-07b3415cdbc4c82cf2204bcb1df28ec65f46cb1e at 07b3415 (+0 new commits): http://git.io/vI8QX
<daviddias>
also known as Mondays? :)
<cryptix>
yea, pretty much :))
<ipfsbot>
[go-ipfs] jbenet deleted feat/httpApiTcpPort0 at 4b337bb: http://git.io/vI8dP
<ipfsbot>
[go-ipfs] jbenet deleted gpe-07b3415cdbc4c82cf2204bcb1df28ec65f46cb1e at 07b3415: http://git.io/vI8d9
<ipfsbot>
[go-ipfs] jbenet deleted gpe-4b337bb7e8786bdab28cc27aa4b6ee8f5dc791a5 at 4b337bb: http://git.io/vI8dP
<ipfsbot>
[go-ipfs] jbenet deleted gpe-efa442ada2b0a709cf98f29d0882fdf9539a2c80 at efa442a: http://git.io/vI8dF
zabirauf_ has quit [Quit: My Mac has gone to sleep. ZZZzzz…]
hellertime has joined #ipfs
inconshr_ has quit [Remote host closed the connection]
<lgierth>
mornin'
<lgierth>
mh now i have a stable username@host and still can't rejoin that +i channel
inconshreveable has joined #ipfs
williamcotton has joined #ipfs
williamcotton has quit [Ping timeout: 252 seconds]
<cryptix>
lgierth: my irc modes stuff is rusty but isnt +i invite only?
inconshreveable has quit [Remote host closed the connection]
<lgierth>
cryptix: yeah and somehow i thought the invitation would stick to user@host combination
<lgierth>
apparantly it doesn't :)
<cryptix>
lgierth: afaik most ircds implement the invite validity only for the connection duration - if setup correctly you can use chanserv to re-request invites on some networks
pfraze has joined #ipfs
www has joined #ipfs
domanic has quit [Ping timeout: 256 seconds]
domanic has joined #ipfs
nessence has joined #ipfs
therealplato has quit [Ping timeout: 246 seconds]
therealplato has joined #ipfs
therealplato has quit [Client Quit]
therealplato has joined #ipfs
antonio_l has joined #ipfs
nessence has quit [Remote host closed the connection]
domanic has quit [Ping timeout: 272 seconds]
ipfs-gitter-bot has joined #ipfs
ipfs-gitter-bot has quit [Remote host closed the connection]
ipfs-gitter-bot has joined #ipfs
ipfs-gitter-bot has quit [Remote host closed the connection]
ipfs-gitter-bot has joined #ipfs
<ipfs-gitter-bot>
(awrelll) testing
<ipfs-gitter-bot>
(awrelll) bridge up !
<ipfs-gitter-bot>
(awrelll) :smile:
<ipfs-gitter-bot>
(awrelll) anyone on IRC?
ipfs-gitter-bot has quit [Remote host closed the connection]
<mmuller_>
yep
ipfs-gitter-bot has joined #ipfs
<ipfs-gitter-bot>
(awrelll) now?
<mmuller_>
yes :-)
awrelll has joined #ipfs
<whyrusleeping>
hrm, gitter again?
<awrelll>
yup
<awrelll>
back online
<ipfs-gitter-bot>
(awrelll) this is a Gitter Message !
<Luzifer>
tired… hoping to be able to release work to open source and public...
<Luzifer>
btw: I threw a few pull-requests after you… ;)
<whyrusleeping>
oh yeah? i'll have to go do some CR
<Luzifer>
in your IRC bot… I was trying to port our radios bot onto your hellabot lib but then I saw that huge pile of s... I would have to migrate too and gave up xD
<Luzifer>
s/IRC bot/IRC bot lib/
<whyrusleeping>
sweet! hellabot is my favorite :)
<Luzifer>
then maybe you'll like my additions… (or you'll hate them :D)
<whyrusleeping>
lol, that code was a bit messy. thanks!
<whyrusleeping>
also, TIL notice is a thing on irc
<Luzifer>
hmmm… okay… irccloud does not show sent notices?
<Luzifer>
O_o
<whyrusleeping>
mine showed up in the server log window
<daviddias>
lol nope, but I'm still waiting for it to buffer to watch the first second, maybe when that finishes I will have a 8K screen
<daviddias>
ahah
<whyrusleeping>
so, 8k is basically four 4k screen stitched together, right?
nessence has quit [Remote host closed the connection]
<Tv`>
whyrusleeping: 4k screens started out as two panels stitched together
<whyrusleeping>
Tv`: i thought it was four 1080p's stitched
<Tv`>
whyrusleeping: so i just offered to rebase the pin branch, but then i realized i don't really know much about the migrations
<Tv`>
whyrusleeping: i only heard about a vertical seam
<whyrusleeping>
hrm... maybe then?
<whyrusleeping>
re: migrations, dont worry about it
<Tv`>
ok thanks
<whyrusleeping>
i'll take of that as need be
williamcotton has quit [Ping timeout: 264 seconds]
<Luzifer>
wow… that 8k video is buckering :D
<Luzifer>
400% CPU… :D
<Luzifer>
6000 RPM on both fans…
<Luzifer>
(but it looked really good on a 1440p screen xD
<lgierth>
jbenet: whyrusleeping: getting rid of ipfs_bootstrapd involves porting the blocklist to ipfs daemon right?
<whyrusleeping>
lgierth: yes
<lgierth>
ack, on it
<whyrusleeping>
currently, theres nothing on it, so it could be just put as a 'todo'
<lgierth>
ah. good one. i didn't actually look at the list
<lgierth>
is it for dmca safe harbour?
<whyrusleeping>
yeap
cjdmax has joined #ipfs
cjdmax_ has joined #ipfs
<daviddias>
Luzifer ahaha it is a power hungry video
<Luzifer>
daviddias: for sure! my macbook retina is dying… xD
* whyrusleeping
turns on his desktop
<daviddias>
maybe Apple will launch today some 8K hw goodness
whyrusleeping has quit [Ping timeout: 272 seconds]
<daviddias>
ok, got what was the problem with the interop problems I was having between go and node multistream impls. It seems that sorribas impl of length-prefixed-message doesn't include a '\n', and since we changed the protocol to include it, well, it barfed :). I'll just add the '\n' myself to the buffer that is passed onto lpm when a write occurs and trim it for
<daviddias>
the reads
whyrusleeping has joined #ipfs
<whyrusleeping>
i'm getting real tired of ramnode about now
<whyrusleeping>
just wish digital ocean didnt cost so much...
<whyrusleeping>
cant justify spending $60/year for an irc bouncer
<whyrusleeping>
hrm... does that say meetings at 5pm?
<lgierth>
ah right, missed the async/irc bit of part 1
<jbenet>
I want to experiment with this today -- the changes are: (a) move social part to "community meetup" (time can be figure out in next couple days) (b) do the checkin asynchronously on irc (feels like there's a bunch of dead time for others when a few people dive into things on hangouts)
<daviddias>
ah ok :) I read it, but didn't understand that we were actually cutting the hangout itself
<jbenet>
and (c) make time for smaller groups to discuss things live. people can just show up to whichever ones they want
<jbenet>
daviddias: yeah, i also like the hangout, am just wary of having too much required sync time that's underutilized?
<jbenet>
the per-topic hangouts may work well instead
<lgierth>
last sprint (my first) i tidied up the gateway infrastructure provisioning, made sure ipfs_bootstrapd processes get restarted regularly. the switch from ipfs_bootstrapd to ipfs daemon might make it into a PR today. discovery via cjdns didn't get finished
<jbenet>
lgierth, great. and havent noticed any more problems with gateways, so reboots doing well.
<lgierth>
coolio
<jbenet>
lgierth it's ok to move to the daemon without the blocklist, as we havent been asked to take anything down yet
<jbenet>
lgierth: but yeah we probably want it in the daemon eventually.
<lgierth>
oh ok. i'll leave it in a branch somewhere
<jbenet>
lgierth: anything we can help with on cjdns front?
<lgierth>
don't think so, i just didn't look into dialing during the sprint and took care of the other tasks instead
<jbenet>
yep sounds good
<jbenet>
i'll make some other issues about gateway/infra today.
<jbenet>
--- before going through the rest of the list: for the per-topic discussions, i figured we could allot 30m to each of them going down the list. i want to frontload all the ones that people in germany may want to be part of, so it doesnt get too late, etc. but maybe:
<whyrusleeping>
jbenet: mars has been up 6 days :)
<jbenet>
(sorry for the turbulence as we try this out-- want to maximize the use of everyone's time and make space for good discussions to happen without bogging down everyone else who doesn't care)
notduncansmith has joined #ipfs
notduncansmith has quit [Read error: Connection reset by peer]
<jbenet>
(if want to attend something and time there doesn't work for you just bring it up and we can readjust)
<jbenet>
Ok, continuing--
<jbenet>
I'll go from the top-- @whyrusleeping sync?
<daviddias>
Sounds good :) Thank you for taking care of all of this people orchestration :)
<whyrusleeping>
one sec
<jbenet>
(daviddias: i'm not very good at real time people orchestration, etc, so if anyone has better ideas / is good at it, lmk!)
<jbenet>
whyrusleeping can come back, i'll skip to next-- @krl ?
<whyrusleeping>
made the switch from util.Key to block.Key, eventually, i'd love to not have the util package
<daviddias>
I have some questions specifically about node-ipfs, but I wouldn't mind sitting through some of the other sessions to soak up a better sense of all that is going on, if possible :)
<whyrusleeping>
patch is implemented, RFM
<whyrusleeping>
i figured out what was going on with the weird network latencies, waiting on a protocol update to fix it
<whyrusleeping>
and i implemented multistream and a helper tool to go along with it
<jbenet>
+1 to removing the util package in favor of more granular things.
<whyrusleeping>
i'm wanting to get multistream integrated into go-ipfs soon
<jbenet>
i thought we merged patch?
<jbenet>
maybe we didnt
<whyrusleeping>
hasnt yet
<whyrusleeping>
oh, wait
<whyrusleeping>
patch is merged
<whyrusleeping>
the bugfix hasnt
<jbenet>
yeah :)
<jbenet>
oh
<jbenet>
right-- tests ?
<whyrusleeping>
yeap
<whyrusleeping>
there there
<whyrusleeping>
^y
<jbenet>
ok great will merge right after this
<whyrusleeping>
EOF
domanic has quit [Ping timeout: 256 seconds]
<jbenet>
sounds good-- krl around?
<jbenet>
else i'll go
<jbenet>
spent most week CRing stuff. also handled wire things (made wire spec thing, still needs improvements) figured out multistream things,
<jbenet>
CRed the S3 stuff -- havent gotten to migration because it still imported all of ipfs
<jbenet>
(whyrusleeping any chance of reducing that now we have the key change?)
<jbenet>
I made https://github.com/jbenet/ipfs-vbox which fixes up that makefile to boot arbitrary isos -- it works nicely, actually. i've been booting linux VMs with it. it's speed is terrible because we still dont fetch dags quickly.
<whyrusleeping>
jbenet: its not importing all of ipfs anymore
<whyrusleeping>
i pruned a bunch out
<jbenet>
ok sounds good, will take a look right after
<jbenet>
and a bunch of random small things (logo, webui, electron, etc)
<jbenet>
(mostly lots of CR)
<jbenet>
ok next
<jbenet>
Tv` -- if you're around, maybe discuss the thoughts on pinning? (still need to read backlog)
<Tv`>
jbenet: rebased the pull request, waiting for it to be merged to dev
<jbenet>
Tv` awesome, will do that. thought on making the last objects?
<Tv`>
"the last"?
<jbenet>
the root pinning objects-- "pins/{local, direct, indirect}" etc
<Tv`>
there's just one object
<jbenet>
oh i missed the update. ok will take a look.
<jbenet>
(had thought they were stored at different keys, or whatever)
<jbenet>
ok sounds good. anything else needed for S3?
<Tv`>
jbenet: there's a top-level object with links
<Tv`>
s3 needs to be rebased after pin is merged
<Tv`>
there's no real work to be done on it until it's clearer what's wanted of it
<jbenet>
ok will take a look at that.
<Tv`>
the branch only moves the datastore, nothing else
<jbenet>
if can start discussion by listing open questions on that PR that would help me page in concerns, etc.
<jbenet>
ok gatesvp around?
<jbenet>
dont think chriscool is around either
<jbenet>
wking!
<chriscool>
Hi everyone
<jbenet>
oh hello
<chriscool>
yeah I am around!
<jbenet>
somehow didn't see your name in my people list. (fail)
<jbenet>
chriscool want to give update?
<chriscool>
no problem
<chriscool>
yeah I was about to add GitCop at least to go-ipfs
<chriscool>
is everyone ok with that?
<jbenet>
sounds good. that will be good to have. I'll make an issue with the remaining "OSS Hygiene" things to do
pfraze has quit [Ping timeout: 256 seconds]
<jbenet>
it will be a bit rocky at the beginning as everyone adjusts to add the commit hooks
<chriscool>
also I had a quick look at circleci IPFS_PATH problem and din't see what's wrong
<chriscool>
yeah but I think it is documented properly howto setup the hook
notduncansmith has joined #ipfs
mildred has quit [Ping timeout: 265 seconds]
<jbenet>
chriscool: one thing that may be useful is to have a script to setup the various things like gitcop, CI, waffle, etc. all the stuff that's "per-repo" though this may be easier for some of the node people around.
notduncansmith has quit [Read error: Connection reset by peer]
<jbenet>
chriscool: yeah i think so too, it should be fine. let's turn gitcop on for go-ipfs
pfraze has joined #ipfs
<chriscool>
I think GitCop can only be setup using the web interface, but its very simple to setup
<jbenet>
chriscool: i'll send an invite to a meldium team (meldium.com) for managing credentials (like gitcop). will put ipfsbot creds there too-- it's what i use to setup CI in places
mildred has joined #ipfs
atrapado has joined #ipfs
<chriscool>
ok great
<jbenet>
ok wking?
compleatang has quit [Read error: Connection reset by peer]
<wking>
jbenet: hi :). I haven't had time to review go-ipfs#1151, since I've been trying to get the Docker-registry backend out the door.
<jbenet>
sounds good, yeah the docker stuff is more important
<wking>
Current issues there are the lack of a patch command in ipfs-shell (whyrusleeping, I can add this if you don't get to it before me)
<chriscool>
when I had a look at circleci I was wondering I would I get credentials
<whyrusleeping>
wking: i have it implemented, waiting on a PR to merge on go-ipfs
<jbenet>
(chriscool: yep! i'll add them to meldium. i use https://twitter.com/ipfsbot because i dont want to give all these CI systems R/W access to all my repos)
<wking>
And the missmatch between our data and the Docker-registry storage API (see issues I referenced 50 minutes ago)
compleatang has joined #ipfs
<wking>
whyrusleeping: ah, thanks :)
<jbenet>
wking: i think changes on docker's end wont happen anytime soon.
<jbenet>
wking: can we work around them?
<wking>
yeah, by just returning Now() for ModTime and storing a path string in the FileInfo
<wking>
So the patch entry in ipfs-shell is the main thing.
<jbenet>
wking: other than that, what's needed for docker? if we want to discuss it with docker folk before dockercon, has to be soon (jun 22-23)
<wking>
whyrusleeping: Can you PR the ipfs-shell patch so I can start writing my consuming code there?
<jbenet>
okay let's get patch into ipfs-shell -- (whyrusleeping can you help with that?)
<lgierth>
(is the plan to have a big bang docker-on-ipfs announcement at dockercon? :P)
<wking>
jbenet: that should be it, I just need to finish wiring the driver together, and I imagine I'll turn up some bugs in my current WIP while I'm testing it
<chriscool>
By the way it looks like I cannot add GitCop to go-ipfs because I don't have the rights to create a webhook
mildred has quit [Ping timeout: 265 seconds]
<jbenet>
lgierth: ideally-- but we'd need to have a damn good tool for that. it's not impossible to get a breakout slot at dockercon if what we have is _amazing_, but anything short of that is probably too late.
<jbenet>
lgierth: and slot or not, we can always announce it concurrently, but again needs something really good.
<lgierth>
yep
<jbenet>
chriscool: i can set it up till you have creds, etc (can use ipfsbot for all this)
<chriscool>
jbenet: yeah ok please do it
<chriscool>
it's easy
<jbenet>
wking ok sounds good! let's make sure we get the patch thing in for you. think it would be reasonable to have a working thing on wed or thurs?
<wking>
I'll try and fit an all-nighter in there ;).
<jbenet>
wking: don't stress, we can do it later. (i'll just be in SF and possible to swing by and chat with docker folk)
<jbenet>
ok
<jbenet>
rht__ around o/
<jbenet>
?
<Tv`>
whyrusleeping: there's lots of "mp" fields with Pinners stored in them (or not) that aren't used
notduncansmith has quit [Read error: Connection reset by peer]
<daviddias>
it can talk with the go-multistream impl, however, there are some bits in the go-impl that need to be added (such as protocol ACK, or listener/broadcaster only mode)
<jbenet>
daviddias: cool! it's a good step to convergence.
<jbenet>
daviddias: the next step is spdy on go-ipfs + node-ipfs. we can discuss this more in the node-ipfs discussion
<daviddias>
I would love to work on the DHT + mDNS Node impl this week, but I will need some guidance/spec on how things are structured. Also, getting spdy streams working on top of MultiStream
<daviddias>
okok :)
<jbenet>
(which we're going into now, but i'll adjust the times)
<jbenet>
yeah the DHT + mDNS stuff will be really useful first steps.
<jbenet>
just getting a node-ipfs dht client will be good.
<jbenet>
(and usable in other modules, etc).
<jbenet>
ok who's not listed in the issue--
<jbenet>
tperson ?
<daviddias>
tperson is referenced (as travisperson)
<jbenet>
think he's not around atm.
<jbenet>
ok am i forgetting anyone? anyone else made it back? -- krl cryptix rht__ gatesvp harlantwood ?
<jbenet>
kbala ?
<whyrusleeping>
wking: i pushed patch
<whyrusleeping>
to ipfs-shell
<jbenet>
ok i'll check in with them individually today/tomorrow.
<wking>
thanks :). Want to merge my ipfs-shell PR too?
<wking>
whyrusleeping: ^
<jbenet>
here's the etherpad for this week: https://etherpad.mozilla.org/Fo3cmIE6jk -- I put last week's stuff at the bottom, so that we dont keep bringing things that are old for no reason. take a look and add issues. what to add here may come up / be decided in the next topic focused discussions
<jbenet>
etherpad will be turned into a sprint issue later tonight.
<jbenet>
i expect a lot of these will be sparse, but i'll be at most. I'll add a hangout link to that issue
<jbenet>
ok, anyone remaining feel free to sync with me as you get here.
<jbenet>
daviddias whyrusleeping does 14:00 PDT (in 30min) for node-ipfs discussion work for you? (discuss getting interop)
<daviddias>
it works for me
<whyrusleeping>
jbenet: sure
<whyrusleeping>
wking: yeah, i can merge that
<whyrusleeping>
boom
<whyrusleeping>
done
<daviddias>
nice! :)
<whyrusleeping>
jbenet: can we merge 1334?
<whyrusleeping>
wking needs it
notduncansmith has joined #ipfs
notduncansmith has quit [Read error: Connection reset by peer]
<dPow>
This sprint change is a little confusing...but I think I've got it. I won't be able to make the later meeting times as I scheduled myself out for the normal time
www has joined #ipfs
<ipfsbot>
[go-ipfs] jbenet closed pull request #1334: 'key' objects dont marshal to json well (master...patch/patch) http://git.io/vIToJ
<jbenet>
dPow: sorry-- thanks for bearing with me changing things up last minute. (i think i confused krl cryptix rht and others too)
<wking>
whyrusleeping: thanks :)
<jbenet>
i think this may be a better flow going fwd if we're all here and ready to go. i think we could compress it to 30-40min instead of 75min.
<wking>
jbenet: and thanks for landing go-ipfs#1334 :)
<jbenet>
chriscool what's the gitcop hook url?
<chriscool>
well I tried to set it up using the web interface and it failed with: "Gitcop could not automatically create your webhook, please create it manually using the following steps"
<chriscool>
Visit the new webhook page on GitHub for this repository: ipfs/go-ipfs
<chriscool>
then they say: Ensure the "Content type" is: application/json
<chriscool>
and Enter the following "Secret": 5ddebe18a8b2037fd3a435afb29fc57f
<chriscool>
Ensure the following events are checked: Push, Pull Request and Pull Request review comment
<chriscool>
Ensure the webhook is active
<chriscool>
Press Add webhook and refresh this page
<chriscool>
tell me when you have pressed Add webhook
<jbenet>
chriscool: ok it seems to be good.
<whyrusleeping>
jbenet: i think i may have found a memory leak...
<chriscool>
there is still: ipfs/go-ipfs Webhook not configured for this repository:
* lgierth
out heading home, will be back on time for infrastructure sync
<jbenet>
lgierth: sgtm
<jbenet>
chriscool: i'll try adding it from the ipfsbot account so that we can manage it from there
<anshukla>
jbenet: are you working out of dtpa today?
<jbenet>
chriscool: want to delete it from your gitcop account? (i.e. so that we can all access it from ipfsbot without giving people's credentials)
<jbenet>
anshukla: am at stanford actually. y2e2.
<chriscool>
ok
<chriscool>
done
<anshukla>
coupa?
<jbenet>
anshukla: sorry my ability to multitask is not strong tofay
<whyrusleeping>
my ability to task is not strong today
<anshukla>
no worries, that's probably a skill. i'll give you a heads up here as I'm on the way.
<chriscool>
jbenet you can add it using the ipfsbot account
<jbenet>
chirscool i'll send you the credentials for ipfsbot via meldium. sec
<whyrusleeping>
daviddias: do go and node multistream interop?
<daviddias>
I've made a small test to make sure they interop (in the sense of one can read and write messages to the other), following your impl - https://github.com/diasdavid/node-multistream/blob/master/tests/go-interop.js . However, we need to add a bit more to the go-impl to full interop, since there are now some protocol acks happening, following
<whyrusleeping>
daviddias: doesnt he say not to add an ACK?
notduncansmith has joined #ipfs
<daviddias>
right, not the "ACK", but the protocol is always sent back too
notduncansmith has quit [Read error: Connection reset by peer]
<jbenet>
chriscool: i've sent you an invite to meldium. should give you the creds for ipfsbot. should be able to login to github (i usually use an incognito window for separate cookies) and thus to gitcop (sorry for destroying the work done on setting the messages up)
<whyrusleeping>
daviddias: fairly sure i do that
cjdmax has quit [Remote host closed the connection]
cjdmax_ has quit [Remote host closed the connection]
<jbenet>
chriscool: ipfsbot has admin rights across a bunch of repos on github.com/ipfs. it's good for setting up CI as well. it will give you circleci access that way too.
<chriscool>
jbenet: great, thanks!
<jbenet>
chriscool: i added the repo to gitcop but it asks me for a commit template etc. want to take a look? should be able to setup the hook now as you had before. (and hook may even be setup already. not sure)
<chriscool>
In case you still want to setup GitCop I sent you an email
<daviddias>
whyrusleeping as I understand from your impl, once connected to the server, the server sends back the multi stream header and waits to receive a protocol to match an handler, it never checks to confirm if the client sends back also multistream header, and when a protocol matches to an handler, it doesn't send back (as an ACK) the same protocol header, it
<daviddias>
only send na if not found
<chriscool>
yeah I will take a look anyway to see if meldium works
<lgierth>
the gateways could host the bot too, high-availability pinbot
<whyrusleeping>
lgierth: i would have to change a bunch of pinbot code to support multiple backends
therealplato has joined #ipfs
<whyrusleeping>
but thats not out of the question
<whyrusleeping>
also, remind me never to eat 'spicy cayenne beef jerky' when its 80 degrees fahrenheit and humid out. I'm sweating so much i can barely see
<lgierth>
:>
<lgierth>
but is it worth it?
<whyrusleeping>
oh yes
<whyrusleeping>
its so good
<lgierth>
:)
notduncansmith has joined #ipfs
notduncansmith has quit [Read error: Connection reset by peer]
tso has joined #ipfs
therealplato has quit [Ping timeout: 244 seconds]
<lgierth>
so, mh, i don't know if i see the whole picture with pinning yet, but i have a few thoughts. pinbot is good for adhoc pins, and general tooling around pinning. pins that are important by some any of importance, are in need of deploy and backup processes within the domain/project they are used by
therealplato has joined #ipfs
<lgierth>
does this sound about right?
<lgierth>
s/some any/any measure/
<whyrusleeping>
yeah, we need a way to ensure certain things are *always* pinned by at least one node (preferably more)
<whyrusleeping>
for example, the webui
<whyrusleeping>
that *has* to be accessible instantly to everyone
<lgierth>
right now the webui seems to be pinned via --assets-path which will just regularly add the directory
<lgierth>
and the stuff in refs-to-seed ^ is pinned by hand?
<whyrusleeping>
refs-to-seed is pinned by hand
<whyrusleeping>
and so is the webui AFAIK
<lgierth>
yeah you're right, setting --assets-path is an artifact of the old way of dploying webui
<wking>
for this sort of pin distribution, we probably want a way to pin an IPNS entry. Then you configure a few nodes to pin ipfs.io/important (and whatever else they want to pin), and the nodes will hear about (somehow) updates to that IPNS path, unpin the old content (or not?), and pin the new content.
<wking>
so pinbot basically turns into a way to patch/update ipfs.io/pin/<name> to point to a given hash
notduncansmith has joined #ipfs
notduncansmith has quit [Read error: Connection reset by peer]
<whyrusleeping>
wking: that's a good idea
<wking>
whyrusleeping: I think it makes sense ;), but maybe it has to wait until after we've worked out pub/sub for registering "tell me about updates to the IPNS record {hash-for-ipfs.io-node}". Do we do that yet?
<whyrusleeping>
nope
<whyrusleeping>
pubsub is down the road a bit, near to the live streaming via dht system
<whyrusleeping>
which will just be a super fancy hack
notduncansmith has joined #ipfs
notduncansmith has quit [Read error: Connection reset by peer]
aatkin_ has joined #ipfs
<aatkin_>
Ran into Error: Cannot find module '/usr/local/bin/ipfs-webui/dev.js' when trying to reinstall webui
<whyrusleeping>
krl: tperson o/
ianopolous has joined #ipfs
<aatkin_>
Does a throw here: module.js:340
aatkin_ has quit [Quit: Leaving]
<jbenet>
lgierth on pinning: yeah we want to use an IPNS entry once its robust, but we're not there yet. i've considered turning https://github.com/ipfs/infrastructure/blob/master/seeding/refs-to-seed into an ipfs object, and now with patch we can, but we want to maybe automate that a bit better.
<jbenet>
lgierth: all these processes require tools for us to build (kind of like all the git porcelain) and so far i've been doing a lot of it manually until the tools are there / to figure out what's the right thing to build.
<jbenet>
(i.e. doing things manually exposes you to all the tricky details up front so you get a sense of what the tool(s) should really do
<jbenet>
Tv` exactly, just human editing an ipfs object isn't as nice right now. ipfs object patch helps, but im strongly doubting the choice of protobuf. msgpack or ubjson would make it so we can make that a json or cson object to be imported trivially.
<jbenet>
cat refs-to-seed.json | ipfs object put --enc json
<Tv`>
*shrug* i would have approached that via the filesystem, myself, i think
<jbenet>
yeah though mountable filesystem is not an option always, and in those, having a human-readable single-file format that is easily importable helps
<Tv`>
i thought this was something updated rarely and the hash embedded as a constant in source code