<jbenet>
whyrusleeping: we've reached bottlenecks with bitswap's perf. i see lots of dupes sent, and huge waits on RTTs. we need the path queries + requesting from peers smarter (i.e. guiding them to send different pieces)
notduncansmith has joined #ipfs
notduncansmith has quit [Read error: Connection reset by peer]
warptangent_ has joined #ipfs
warptangent_ has quit [Changing host]
<whyrusleeping>
jbenet: yeap, i know. its the first thing on my last after dev0.4.0 merges
<whyrusleeping>
i dont want to touch it until then though because theres lots of things in dev0.4.0 that change with respect to my perf work
rjeli has quit [*.net *.split]
warptangent has quit [*.net *.split]
machrider has quit [*.net *.split]
step21 has quit [*.net *.split]
silotis has quit [*.net *.split]
gozala has quit [*.net *.split]
oleavr has quit [*.net *.split]
dandroid has quit [*.net *.split]
dandroid_ is now known as dandroid
oleavr_ is now known as oleavr
rjeli has joined #ipfs
gozala has joined #ipfs
machrider has joined #ipfs
step21_ has joined #ipfs
step21_ is now known as step21
deltab_ has joined #ipfs
[epitron] has joined #ipfs
notduncansmith has joined #ipfs
notduncansmith has quit [Read error: Connection reset by peer]
<richardlitt>
daviddias: saw! Will be here 2:30 EST (in four hours).
Encrypt has quit [Quit: Quitte]
akhavr has joined #ipfs
notduncansmith has joined #ipfs
notduncansmith has quit [Read error: Connection reset by peer]
ygrek has quit [Ping timeout: 268 seconds]
border_ has joined #ipfs
border has quit [Ping timeout: 252 seconds]
bedeho has joined #ipfs
mildred has quit [Ping timeout: 265 seconds]
notduncansmith has joined #ipfs
notduncansmith has quit [Read error: Connection reset by peer]
marianoguerra has quit [Ping timeout: 250 seconds]
jamescarlyle has joined #ipfs
jamescarlyle has quit [Ping timeout: 272 seconds]
notduncansmith has joined #ipfs
notduncansmith has quit [Read error: Connection reset by peer]
notduncansmith has joined #ipfs
notduncansmith has quit [Read error: Connection reset by peer]
border has joined #ipfs
notduncansmith has joined #ipfs
notduncansmith has quit [Read error: Connection reset by peer]
Gaboose has joined #ipfs
<Gaboose>
there's a gateway from http to access ipfs, what about one from ipfs to access http files?
<Gaboose>
is it possible to make like an ipfs node that dynamically advertises itself on the DHT, when a particular block is requested?
<Gaboose>
and then serves them from arbitrary sources
jamescarlyle has joined #ipfs
<Gaboose>
or does the DHT not work that way
jamescarlyle has quit [Remote host closed the connection]
jamescarlyle has joined #ipfs
notduncansmith has joined #ipfs
notduncansmith has quit [Read error: Connection reset by peer]
jamescarlyle has quit [Remote host closed the connection]
<daviddias>
Gaboose: IPFS nodes that are serving a set of blocks will 'provide' to the DHT the information that they are serving those blocks
<daviddias>
once a block search is performed, the 'Record' of where the block is being provided, is retrieved through the DHT so that the peer looking for the block knows which peer to ask the block for
jamescarlyle has joined #ipfs
notduncansmith has joined #ipfs
notduncansmith has quit [Read error: Connection reset by peer]
<krl>
I'd like to finally get back into the web app things now, and set up the webcomponent examples. A candidate could be ipfs-add-file, which i already have in starlog, a component that provides a drag-drop or click-file-select interface, and just fires the hash of the added file as an event.
Leer10 has joined #ipfs
<jbenet>
krl: does the starlog work fully?
<krl>
there's ui issues, and you cannot update your posts. it's demo ready but not really ready for serious use
jamescarlyle has joined #ipfs
<jbenet>
krl: link doesn't work. i get "Uncaught TypeError: Cannot read property 'type' of undefined" in the console.
<krl>
in chrome or firefox?
<jbenet>
krl: and when i do "ipfs name resolve Qmbr5g2KAKj8yKFtg6qeQ9wtyg9tDbTjURugf5pNiMjcMP" in the cli, resolution fails.
<jbenet>
chrome
<jbenet>
ipns is not ready, this has to work with just ipfs for now, updating the url. so you can at least hand people a link, even if the head keeps changing.
<krl>
yes, this is how it's supposed to work now
<jbenet>
ah so Qmbr5g2KAKj8yKFtg6qeQ9wtyg9tDbTjURugf5pNiMjcMP is not a name then?
notduncansmith has joined #ipfs
notduncansmith has quit [Read error: Connection reset by peer]
<jbenet>
(it would be useful to use the proper roots, /ipfs/Qmbr5g2KAKj8yKFtg6qeQ9wtyg9tDbTjURugf5pNiMjcMP and /ipns/... to be able to tell)
<jbenet>
i know /starlog looks cool, but if you want it there, then do /starlog/<full-path> -- /starlog/ipfs/Qmbr5g2KAKj8yKFtg6qeQ9wtyg9tDbTjURugf5pNiMjcMP or /starlog/ipns/Qmbr5g2KAKj8yKFtg6qeQ9wtyg9tDbTjURugf5pNiMjcMP
<krl>
yes, i just did not want to make the url longer
<jbenet>
anyway discuss later probably.
<krl>
k
<jbenet>
lgierth ?
<lgierth>
ok next?
<lgierth>
yeah
<lgierth>
- had a really slow week, but a few nice things got finished anyhow, and i'm happy to largely leave the infrastructure be for now.
<lgierth>
- done:
<lgierth>
- fixed CORS for https:// on the gateways, as per @jbenet's bug report
<lgierth>
- got another SSL cert, for testing on dogfeed.ipfs.io
<lgierth>
- switched over to ipfs.io, with gateway.ipfs.io being still in place, with no redirect so far. we need to talk about security implications re: ipfs.io
<lgierth>
- made /ipfs,/ipfs,/api work alongside an ipfs-hosted static page
<lgierth>
- not so completely done:
<lgierth>
- denylist -- another week without much progress, but we know what to do know and i'm working on it pretty much right now
<lgierth>
- get another person to deploy -- there are two lined up just waiting for a good occasion :)
<lgierth>
- peer discovery using cjdns -- i might other babysteps progress on go-cjdns though
<lgierth>
- nothing:
<lgierth>
- custom static pages -- postponed that for now
<jbenet>
(ah sorry was prompting you to ask the next person-- maybe there's lag. anyway :D)
<jbenet>
lgierth: TLS stuff is great!
<lgierth>
yeah and it works well so far. please tell me if you see anything weird like the issue you posted w/ chrome
jamescarlyle has quit [Ping timeout: 256 seconds]
<jbenet>
i'm happy to deploy later tonight
<lgierth>
coolio, i'll be here. not for much longer after 11pm UTC, but anyhow
<daviddias>
sure, let me just finish writing my update
<lgierth>
whyrusleeping: mappum: you there?
<mappum>
i'm here
<mappum>
oh, it's monday D:
<lgierth>
yeah! :)
warner` is now known as warner
<whyrusleeping>
lgierth: ping
<whyrusleeping>
i wont be super available until later today, three day weekend traveling and such
<lgierth>
whyrusleeping: ok no problem! if you have anything to post, feel free -- otherwise just post to the pm issue
<whyrusleeping>
lgierth: sure thing. I'll write that up
<whyrusleeping>
(also forgot it was monday... my bad)
<lgierth>
daviddias: yell "here" when you wanna post
<jbenet>
whyrusleeping maybe post alternate times that work for you for discussions on the pm issue
<jbenet>
(late today or tomorrow)
<lgierth>
jbenet: anything big on the horizon that we should pay attention to this week?
<whyrusleeping>
lgierth: container.camp
<whyrusleeping>
jbenet: will do
<jbenet>
lgierth: i'm focused on starship -- the containers/ipfs tool i'll talk about at container.camp and I wont have time for much else. i've help from others on that.
<jbenet>
not sure if there's another general focus-- i think people have their pieces cut out for them?
<lgierth>
yep cool -- i do at least
<daviddias>
lgierth: thing I got it :)
* daviddias
here
<lgierth>
\o/
<daviddias>
s/thing/think
<daviddias>
During last week, in a overview, managed to develop more of the Records system that libp2p uses, once me and @jbenet managed to figure how to push IPLD forward. Excited for the WebRTC wg F2F meeting is this Wednesday, where I'll be attending remotely.
<daviddias>
- [x] Decide on minimum viable IPLD, that doesn't break json-ld and can be used for IPFS with @jbenet
<daviddias>
- [ ] {Still working on it} kad-record-store
<jbenet>
daviddias that's awesome
<daviddias>
Thank you jbenet :)
<jbenet>
looking fwd to playing with it
<lgierth>
oh wow ietf working group? cool! -- juliusz made it sound really scary in maribor ;)
<daviddias>
lgierth: it is a W3C WG, but yes, there are huge ties with the IETF RTCWeb WG
<lgierth>
there's a webrtc and an rtcweb working group? :/
<lgierth>
damn you, committee bodies!
<daviddias>
kad-record-store is the last building block to have a libp2p that offers the same features (and a lil more) that the go-ipfs network stack has
<daviddias>
lgierth: yes, W3C handles API level stuff, while IETF handles protocol level stuff
<lgierth>
i see, ok
<lgierth>
daviddias: what more do we need for good ipns, in addition to record stores and record validity?
<jbenet>
we need to implement all that in go-ipfs
<jbenet>
(i can discuss that in my update)
<lgierth>
yep -- cool
<lgierth>
wanna go?
<daviddias>
lgierth: I honestly haven't thought that thoroughly as I've been developing this with the 'provider records in mind', nevertheless I've been following the IPRS spec which has several types of records in mind
<lgierth>
got it
<daviddias>
lgierth: also what jbenet , a good next step is to start stripping out the pieces that make the network stack to a libp2p go module thing, so that I can make both implementations match up and ensure they are fully compatible (and compliant with the libp2p spec)
<daviddias>
but yes, kad-record-store is like the last building block to have a libp2p thing going, woo! :D such excite to make that available
<lgierth>
:):)
<daviddias>
then it is all about those extra features + code coverage
<lgierth>
coolio
<jbenet>
daviddias: yeah shipping libp2p will be a big deal
<whyrusleeping>
i'm ready whenever
<jbenet>
ok i'll push my update
<lgierth>
whyrusleeping: fire
<jbenet>
(ok whyrusleeping go, race conditions!)
<whyrusleeping>
- [x] implement symlinks in unixfs
<whyrusleeping>
- [x] print panics when they happen in the API
<whyrusleeping>
- [x] rebase dev0.4.0 again
<whyrusleeping>
- [ ] get mfs merged
<whyrusleeping>
- change of plans, going to rework
<whyrusleeping>
- [x] commands lib can now return errors correctly from a goroutine through the channel marshaler
<whyrusleeping>
- [x] blockservice no longer ignore blockstore errors
<whyrusleeping>
- [x] gx-go-import works and can import go-ipfs (albeit not ideally)
<whyrusleeping>
(i probably missed something, github makes it difficult to look back on your week)
<jbenet>
whyrusleeping you can click on each day on your grid: github.com/whyrusleeping
<whyrusleeping>
whoa
<whyrusleeping>
TIL
<mappum>
or change the Period dropdown to 1 week
<jbenet>
whyrusleeping: sgtm good stuff fixing all those bugs, and implementing the symlinks
<jbenet>
whyrusleeping want to start an issue tracking all the changed for 0.3.8 so far? may want to cut it soon
<whyrusleeping>
jbenet: yeah, i can work on that
<whyrusleeping>
i would love to try and commit to a release every two weeks
<jbenet>
whyrusleeping: sgtm.
<lgierth>
+1
<lgierth>
and increment version segments purely based on what got merged, instead of saying "let's make a minor release, let's see what we got"?
<whyrusleeping>
yeap
<jbenet>
lgierth: we dont do semver on go-ipfs yet.
border has quit [Ping timeout: 260 seconds]
<jbenet>
we'd be at 10000.0.0 by now if we did.
<lgierth>
ok
<lgierth>
what does 0.4.0 mean then compared to 0.3.x?
<daviddias>
and that would be... over 9000!
<lgierth>
:)
<daviddias>
:D
<whyrusleeping>
that would be an impressive feat seeing as we only have around 4000 commits
<whyrusleeping>
:P
<whyrusleeping>
lgierth: 0.4.0 is a protocol incompatible change
<lgierth>
ok
<lgierth>
cool cool -- jbenet?
<jbenet>
ok
<jbenet>
- this week a lot of my time was taken up by talks in Basel and Zurich. which went well.
<jbenet>
- But putting that on hold as i'm focusing on starship exclusively this week. more on that later.
<jbenet>
- We ran into more snags in the ipld definitions, but seems settled enough to implement in node + go. (thanks @daviddias and @mildred). I've implemented multicodec and everything we need from IPLD (except one last format sync i'll looking at daviddias's module).
<jbenet>
- [ ] IPLD
<jbenet>
- [x] plan fwd with @mildred and @diasdavid
<jbenet>
- [x] made a test for mappum's resolve PR
<jbenet>
- comms
<jbenet>
- [x] talk at ETH (in Basel) to biotech people
<jbenet>
- [x] discussions with a number of swiss crypto folks
<jbenet>
- [x] talk at Webilea (web dev meetup in Basel)
inconshreveable has joined #ipfs
<daviddias>
so much awesome and rich updates and I meant that across the board! :) nice!
<lgierth>
yeah!
<daviddias>
jbenet: yeah, I've been tapping on the latest of those discussions, looking forward to learn about what are the latest conclusions, I believe it isn't anything like a 180 degrees turn
notduncansmith has quit [Read error: Connection reset by peer]
border_ has quit [Ping timeout: 272 seconds]
thomasreggi has joined #ipfs
<jbenet>
hey krl: yay it loaded locally! \o
<jbenet>
\o/
<lgierth>
is that multikey thing online somewhere?
<krl>
i'm sure of the datastructures for scroll+search now, ui is still lacking
<lgierth>
what's the issue with it not loading on ipfs.io? i don't see errors or failed requests
<krl>
lgierth: i'd love for you (or someone who knows the gateways) to look at that
<krl>
lgieth: huuuh, now it loaded in gateway forme
<jbenet>
krl: the link icons on the top-right dont work as expected?
<krl>
lgierth
<krl>
jbenet: ok, another regression :(
<lgierth>
krl: without /latest in the end it loads more stuff, including a 400 response for /ipfs/{{icon}}, and TypeError: this.$ is undefined
<krl>
this is my first polymer project, and the logic probably needs to be rewritten
jhulten has joined #ipfs
<jbenet>
ok krl, let's jump into discussing it in 10min? (brb).
<jbenet>
it may be good, krl, if you could give a quick screenshare demo of it and how it works to us. would help us help you fix things and see where it's headed.
<krl>
i'm not sure how to setup screensharing, i'll research
<jbenet>
thanks krl, looking good. good touch with converting brewster's blog -- and datastruture looks good. (write it up so we can discuss it in the future)
<daviddias>
richardlitt: lgierth posted here the link too
<krl>
brb foods
<richardlitt>
Keep getting this: This party is over... But you can start a new one. Error: HJR: 2-RNF
<krl>
or bbl
<jbenet>
krl: the UI/UX polish is critical else nobody can use it, so maybe let's focus on that. can maybe get others' help (this may be a good project for richardlitt actually, we can discuss later)
jamescarlyle has quit [Remote host closed the connection]
<krl>
richardlitt: o/
<richardlitt>
Word. All I need is clear issues and I can help out with that.
<jbenet>
lgierth: i cant join. maybe hangouts arent forever
border_ has quit [Read error: Connection reset by peer]
<jbenet>
lgierth: same err as richardlitt
<lgierth>
the hangout should be open to anyone noe
notduncansmith has quit [Read error: Connection reset by peer]
deltab has quit [Ping timeout: 250 seconds]
marianoguerra has quit [Ping timeout: 246 seconds]
inconshreveable has quit [Read error: Connection reset by peer]
inconshreveable has joined #ipfs
<lgierth>
oh hey, i propose IPE as the name for cjdns-in-go
<lgierth>
interplanetary express
<lgierth>
various subsets of it can be plugged into libp2p
<richardlitt>
IPE passes wordsafety.com. Sounds good to me.
<ipfsbot>
[go-ipfs] mappum pushed 2 new commits to ipfs-path-resolve: http://git.io/vZLF4
<ipfsbot>
go-ipfs/ipfs-path-resolve 91c51e9 Matt Bell: Consolidated 'resolve' tests...
<ipfsbot>
go-ipfs/ipfs-path-resolve 5ac7cf9 Matt Bell: Use core path resolver for 'resolve' command...
notduncansmith has joined #ipfs
notduncansmith has quit [Read error: Connection reset by peer]
thomasreggi has quit []
jamescarlyle has joined #ipfs
mquandalle has joined #ipfs
notduncansmith has joined #ipfs
notduncansmith has quit [Read error: Connection reset by peer]
<lgierth>
richardlitt: oh hey thanks for that link
<richardlitt>
Yeah. Cool project.
<lgierth>
maybe IPEx is even better a name
pfraze has joined #ipfs
jamescarlyle has quit [Ping timeout: 246 seconds]
notduncansmith has joined #ipfs
notduncansmith has quit [Read error: Connection reset by peer]
Tv` has joined #ipfs
<Leer10>
How are supernodes involved?
notduncansmith has joined #ipfs
notduncansmith has quit [Read error: Connection reset by peer]
deltab has joined #ipfs
inconshreveable has quit [Remote host closed the connection]
<daviddias>
lgierth: IPE seems like a good catchy name, but why would a different language implementation have a different name? Aren't they going to be compatible?
<lgierth>
cjdns is just such a horrible name
<lgierth>
and cjd should feel bad for choosing it -- although i know he doesn't
<daviddias>
ahahaah
<daviddias>
so, you are doing a fork of it, right?
<lgierth>
an alternative implementation with the goals of 1) libp2p integration and 2) coming up with a spec
<lgierth>
cjd still maintains cjdns, but more scratching his own itch than driving it forward (which is totally cool)
<daviddias>
that is what I was going to ask next, seems like a good opportunity to do a 'spec' for implementations
<lgierth>
it's super important to get this written down
<lgierth>
cryptoauth, the switch, the router, ipv6, iptunnel
simonv3 has quit [Quit: Connection closed for inactivity]
<lgierth>
also makes it easier for others to criticize it, instead of having to fall back to resentment :)
notduncansmith has joined #ipfs
notduncansmith has quit [Read error: Connection reset by peer]
<daviddias>
Those seem to be valid points. It would be good to not fragment the cjdns community though
<daviddias>
is there a proposal for this fork somewhere?
Encrypt has quit [Quit: Sleeping time!]
<lgierth>
not written down...
<lgierth>
but there's the express wish of the relevant people for someone to make it happen :)
<daviddias>
Some people will certainly be against it, which will be ok, but at least it can be a community effort than people thinking just 'someone went rogue'
<lgierth>
the past years everything has been welcomed pretty much. if someone builds something, however obscure, it's cool
<daviddias>
lgierth: I would strongly suggest to document that idea as something you want to happen and will make happen
<lgierth>
yeah that's good advice
<lgierth>
make it visible to everybody
<lgierth>
irc is ephemeral
<daviddias>
a github issue on the project itself is a good medium for discussion
<daviddias>
and you can then always link to it in the newer project