<Evermore>
strangely the directory did not get any bigger when I copied them in
<Evermore>
And not even like in an ipfs way, this is on the disk
Wallacoloo has quit [Quit: Leaving.]
hellertime has joined #ipfs
<Blame>
jbenet: problem sorted. It was on my end.
Wallacoloo has joined #ipfs
inconshreveable has joined #ipfs
<whyrusleeping>
Evermore: thats really sweet :D
notduncansmith has joined #ipfs
notduncansmith has quit [Read error: Connection reset by peer]
<whyrusleeping>
Evermore: yeah, you were probably just changing links around, which is cheap
<Evermore>
whyrusleeping: I don't get it though, I did "cp -r"
<Evermore>
it's ext4, did ext4 add a COW feature or something?
<zignig>
MOO!
<Evermore>
Oh I think I see the problem. Small dir was 189 MB, big one is 475 (now)
<Evermore>
So it's closer to 800 MB, I just didn't read enough sig figs
<Evermore>
700... whatever
inconshreveable has quit [Ping timeout: 256 seconds]
<whyrusleeping>
Evermore: uhm... no. ext4 didnt manage to get cowing afaik
<whyrusleeping>
oh, i see
<Evermore>
It could also be that a lot of the files were duplicated so some of the 189 MB overwrote some of the 475 MB
<whyrusleeping>
ah, could be
* whyrusleeping
wonders if jbenet is gonna have the record spec done soon
notduncansmith has joined #ipfs
notduncansmith has quit [Read error: Connection reset by peer]
daviddias has quit [Ping timeout: 272 seconds]
<headbite>
was away playing trine with my niece and nephew... those latest links look to work... I'll try pinning it locally later and go offline and see if it still works.
notduncansmith has joined #ipfs
notduncansmith has quit [Read error: Connection reset by peer]
daviddias has joined #ipfs
daviddias has quit [Ping timeout: 240 seconds]
daviddias has joined #ipfs
notduncansmith has joined #ipfs
notduncansmith has quit [Read error: Connection reset by peer]
tilgovi has quit [Ping timeout: 256 seconds]
pfraze_ has joined #ipfs
notduncansmith has joined #ipfs
notduncansmith has quit [Read error: Connection reset by peer]
hellertime has quit [Quit: Leaving.]
nessence has quit [Remote host closed the connection]
notduncansmith has joined #ipfs
EricJ2190_ has quit [Ping timeout: 264 seconds]
notduncansmith has quit [Read error: Connection reset by peer]
ei-slackbot-ipfs has quit [Remote host closed the connection]
ei-slackbot-ipfs has joined #ipfs
notduncansmith has joined #ipfs
notduncansmith has quit [Read error: Connection reset by peer]
notduncansmith has joined #ipfs
notduncansmith has quit [Read error: Connection reset by peer]
therealplato has quit [Ping timeout: 264 seconds]
notduncansmith has joined #ipfs
notduncansmith has quit [Read error: Connection reset by peer]
daviddias has quit [Max SendQ exceeded]
Wallacoloo has quit [Quit: Leaving.]
daviddias has joined #ipfs
notduncansmith has joined #ipfs
notduncansmith has quit [Read error: Connection reset by peer]
notduncansmith has joined #ipfs
notduncansmith has quit [Read error: Connection reset by peer]
kyledrake has joined #ipfs
mquandalle has quit [Quit: Connection closed for inactivity]
sharky has quit [Ping timeout: 250 seconds]
notduncansmith has joined #ipfs
notduncansmith has quit [Read error: Connection reset by peer]
sharky has joined #ipfs
pfraze_ has quit [Remote host closed the connection]
Wallacoloo has joined #ipfs
Evermore has quit [Ping timeout: 240 seconds]
nemik has quit [Ping timeout: 276 seconds]
nemik has joined #ipfs
Evermore has joined #ipfs
notduncansmith has joined #ipfs
notduncansmith has quit [Read error: Connection reset by peer]
<jbenet>
Evermore: omg -- sbemail on ipfs. young jbenet is so excited right now.
<Evermore>
now strong bad will never stop checking emails
<jbenet>
whyrusleeping: the last part in the records spec is the object format-- so records are now ipfs objects, and it makes sense to have them reference other things. it makes sense too to make the signatures objects as well.
<jbenet>
(it'll make more sense when you read it-- i'll push drafts in a bit)
<whyrusleeping>
"in a bit" ~ jbenet a month ago
<whyrusleeping>
:P
anshukla has joined #ipfs
zabirauf has joined #ipfs
<jbenet>
whyrusleeping: :P
notduncansmith has joined #ipfs
notduncansmith has quit [Read error: Connection reset by peer]
<krl>
so, i need some CR love on node-ipfs-api and node-ipfsd-ctl
mildred has joined #ipfs
<whyrusleeping>
krl: gmornin'
Blame has joined #ipfs
<krl>
whyrusleeping: mornings!
<whyrusleeping>
do you guys practice that whole daylight savings time thing?
<krl>
whyrusleeping: i did when i worked in an office
<whyrusleeping>
lol, so its an official thing in germany?
notduncansmith has joined #ipfs
notduncansmith has quit [Read error: Connection reset by peer]
<krl>
yeps
<krl>
most of europe i would guess
<whyrusleeping>
okay
<crest>
whyrusleeping: yes every clock jumps an hour twice per year
<crest>
but the PTB (Physikalisch-Technische Bundesanstalt) operates a 77,5kHz radio time service in frankfurt with ca. 2000km range
<crest>
and most radio clocks extract the daylight saving crap from its signal
* crest
keeps his system clocks in UTC and sets TZ in /etc/login.conf
<crest>
even windows can be tough to run the RTC in UTC instead of local time
domanic has quit [Ping timeout: 252 seconds]
daviddias has quit [Quit: My MacBook Pro has gone to sleep. ZZZzzz…]
<whyrusleeping>
crest: interesting...
notduncansmith has joined #ipfs
notduncansmith has quit [Read error: Connection reset by peer]
daviddias has joined #ipfs
mildred has quit [Ping timeout: 276 seconds]
notduncansmith has joined #ipfs
notduncansmith has quit [Read error: Connection reset by peer]
daviddias has quit [Read error: No route to host]
daviddia_ has joined #ipfs
cjb has joined #ipfs
notduncansmith has joined #ipfs
notduncansmith has quit [Read error: Connection reset by peer]
<whyrusleeping>
alright, is anyone abusing the DHT?
PayasR has joined #ipfs
[0__0] has quit [Remote host closed the connection]
<PayasR>
Hi!
<whyrusleeping>
PayasR: heyo
<PayasR>
I was just thinking.. Since linux repos are still stuck with golang 1.3, can't we make a complete installable for Linux that includes golang 1.4+?
[0__0] has joined #ipfs
<PayasR>
Debian repos (apt) to be specific
<whyrusleeping>
PayasR: we compile with 1.3 just fine too
tilgovi has joined #ipfs
notduncansmith has joined #ipfs
notduncansmith has quit [Read error: Connection reset by peer]
<whyrusleeping>
and honestly, we might be able to build on 1.2 now, i dont know if we are using the sync.Pool thing anymore
<PayasR>
Then please update readme. It says you need 1.4+
<whyrusleeping>
huh, i wonder who changed that
<PayasR>
Why add unnecessary headache for the users
<PayasR>
Github issue again?
<whyrusleeping>
yeah, i'd go ahead and file that one
<whyrusleeping>
thanks!
<PayasR>
:)
<PayasR>
I've been talking to people about this.. Some more may join the effort if everything goes well :)
<whyrusleeping>
sweet
<whyrusleeping>
the more the merrier
<PayasR>
And being a version controlled system, do we have something like git diff?
<whyrusleeping>
version controlled system?
<whyrusleeping>
ipfs uses merkledags like git does
<whyrusleeping>
but we dont have the concept of version control built in
lgierth has joined #ipfs
<PayasR>
Okay..
Tv` has joined #ipfs
<jbenet>
PayasR: we will though, have version control objects and tooling
payasrajan has joined #ipfs
<whyrusleeping>
jbenet: question, if bitswap receives a block it already has, should it drop it right there?
nessence has quit [Remote host closed the connection]
<whyrusleeping>
jbenet: i think that most of the NAT issue i've seen recently could be solved with a simple connection request system, where peer A can request a connection from peer B through peer C, where C is already connected to B
tilgovi has quit [Ping timeout: 258 seconds]
<ipfsbot>
[go-ipfs] whyrusleeping force-pushed feat/bitswap-speed from 9bf678c to db8af56: http://git.io/vk0E0
<ipfsbot>
go-ipfs/feat/bitswap-speed 43dd49f Jeromy: Move findproviders out of main block request path...
<ipfsbot>
go-ipfs/feat/bitswap-speed 7ed6436 Jeromy: clean up organization of receivemessage and fix race
<ogd>
jbenet: osbridge is like the community version of oscon
<ogd>
jbenet: so tickets are $300 of $3000 and the talks are actually better
<jbenet>
ogd: nice lgtm
<jbenet>
krl: what else should i TAL at?
<jbenet>
we should have a "needs CR" feed.
<ogd>
jbenet: twice in the last few days i was talking to random peeps about p2p stuff and they were like 'you should check out IPFS' lol
notduncansmith has joined #ipfs
notduncansmith has quit [Read error: Connection reset by peer]
<jbenet>
ogd: \o/ yay
<jbenet>
ogd: btw-- hopefully sometime this summer can talk about some data things, we should finish up that spec we started a long time ago to define research artifacts. anything we can help with on the dat front?
tilgovi has joined #ipfs
<ogd>
jbenet: we are trying to get the beta next week, im finalizing the cli now. then we need to focus on some scientific use cases for a bit and build some demos
<mitchty>
out of curiosity would it be possible to put mutation capable git repositories in ipfs yet?
<jbenet>
ogd: awesome
<jbenet>
mitchty: yeah you can, but ipns is not yet fully robust
<jbenet>
mitchty: so YMMV atm.
<jbenet>
mitchty: try it with "ipfs name publish <hash of repo>"
<mitchty>
jbenet: k last time i tried it wasn't yet up to the task
<whyrusleeping>
jbenet: daggernet
<mitchty>
so i can then say git clone and update on say a fuse fs?
<jbenet>
mitchty yeah it hasn't gotten better yet. also the current git onboarding we have is a bit of a hack. cjb pointed out we can do protocol handlers on git itself (he has a cool gitswarm hack too)
<mitchty>
my goal is to get my repos syncing silly easy (have a backup git:// server I can push to at once so no bigs, i'm testing)
<whyrusleeping>
jbenet: also, waffle.io has a CR feed
<jbenet>
whyrusleeping: daggernet sounds awesome -- neal stephenson awesome -- but it's not a "nice and friendly" name, it sounds hostile.
<cjb>
whyrusleeping: that doesn't stop other people claiming the same username, or give you a path from username to key
<cjb>
I'm caring about providing a decentralized version of the "github.com/user/repo" workflow, where you reach the right user from just their username
<jbenet>
cjb: names need some for of consensus, but relying on PoW blockchain concensus for it seems very wasteful to me. you could have a few different approaches to naming, from federated naming (use email or just DNS or something) to federated consensus (SCP) to web of trust naming (this is not perfect because names are paths, not unique
kyledrake has joined #ipfs
<jbenet>
cjb: i'd use email.
<cjb>
jbenet: if I hadn't signed up to decentralize github, I'd agree, but github supports a very compact username syntax so I wanted to too. Agree that email is nice.
<jbenet>
yeah "@jbenet" is nice.
<cjb>
(I don't really depend on bitcoin; bitcoin provides a mapping from username to hash-of-pubkey, and you can just use the hash-of-pubkey directly after receiving it any other way too.)
<mitchty>
i might be thinking of gitchain, but it had somethign like being able to git clone blah://user/project
<jbenet>
cjb: right, but why not keybase or twitter or github itself? i think using bitcoin for this is like depending on a saturn V to get to the corner store.
<jbenet>
(which _works_ for sure, so maybe keep using it cause it already works)
<jbenet>
but it seems very wasteful to me
<cjb>
mitchty: gitchain definitely used bitcoin :)
notduncansmith has joined #ipfs
notduncansmith has quit [Read error: Connection reset by peer]
chriscool has joined #ipfs
<ogd>
cjb: awesome blog post!
ei-slackbot-ipfs has quit [Remote host closed the connection]
ei-slackbot-ipfs has joined #ipfs
<mitchty>
cjb: was confused a bit on namespaces but it makes some sense, but there still seems like it would be a bit goldrushy in the same vein as twitter in that those that got in first get their names and can squat others etc...
lgierth has joined #ipfs
<ogd>
cjb: also remember never to read hacker news comments
<bedeho2>
jbenet: BitCoin as a whole is very costly, but marginal cost of having 5K machines store 40b per such gittorrent entry doesnt seem *that* socially costly really, given the cost of disk space today. How would the email solution work?
<grawity>
I think it's the "proof-of-work" part that's costly, not the "blockchain" part
Wallacoloo has joined #ipfs
<jbenet>
bedeho2 it's tiny marginal costs today instead of working a bit harder for a better more sustainable solution (like solar or nuclear) that add up to dependence on unnecessary things (like coal or bitcoin)
<mitchty>
i'm curious if gpg public keys could be used as well, that way signed commits could be verified
<mitchty>
or tags at least
chriscool has quit [Read error: Connection reset by peer]
<jbenet>
ogd: im curious, why do you prefer protobuf over msgpack? cause it's smaller? (keys don't have the text?)
<ogd>
jbenet: yea thats the main one. we didnt need self describing
chriscool has joined #ipfs
<cjb>
mitchty: it does at least cost $0.08 per username, which has better squatting characteristics than most centralized sites!
<jbenet>
ogd: i think there's an interesting hack where a stream can include a reference to the spec when it's opened, and every message doesn't need to.
<cjb>
($0.16 per username if you want to do the two-step registration to avoid racing)
<jbenet>
ogd: json-ld -esque.
<ogd>
jbenet: ah yea i can see that
<mitchty>
cjb: true i'm just thinking of squatters taking over certain namespaces is all really, aka google/microsoft/etc....
<bedeho2>
grawity: yes, but cost of actual mining is quite independent of size of blocks, so the real cost is having everyone do the mining at all - to which additional txes add no marginal cost, and then the storage/transmission, where the extra cost is nonzero, but quite low.
<jbenet>
ogd: the problem i keep having with protobuf is it doesn't nest streams well. streams, self-description, etc. are all after thoughts and clunky. i'm on the verge of writing a format called "Linked Serialized Data", which is effectively protobuf + pbs + self description in the stream.
<jbenet>
ogd: well i want something _simepler_ than coercing protobuf to do what i want it to.
<jbenet>
anyway, prob not worth wasting time on
<ogd>
jbenet: seems like msgpack would be more flexible for adding metadata in since you can put any json-like structure you want into it without needing to worry about juggling a bunch of schemas
<jbenet>
ogd: yeah precisely. and the schema can be described in msgpack itself (like json-ld)
<jbenet>
err that is if the "schema saving trick" is used.
<jbenet>
(msgpack puts the schema on every message, but i'm effectively saying, the first object (or a referenced but not necessarily transmitteed object) has the schema and the other message use integer keys like protobuf
<jbenet>
like the ability to not say "my super long key" on every message ever is useful.
<jbenet>
but then again, gzip.
<ogd>
jbenet: haha
nessence has quit [Read error: Connection reset by peer]
notduncansmith has joined #ipfs
notduncansmith has quit [Read error: Connection reset by peer]
<PayasR>
i have initialized cpp-multihash repo
<PayasR>
work should reach some stage in a day or two
<PayasR>
need to port code from openssl/opencrypto libs
<PayasR>
CMake or GNU make.. any comments?
<whyrusleeping>
PayasR: personally i like GNU make
<jbenet>
gnu make for sure.
<mmuller_>
GNU make: defacto standard, use autotools and everyone will know what to do. CMake - unlike autotools, is actually sane. Does require you to grok CMake.
<PayasR>
same pinch. but GNU make would bind the code too much to the GNU/Linux/UNIX based systems
<jbenet>
krl: who was the guy at the end2end meetup that was asking about threat models? i want to tag him on a conversation about the record system so he gets what i meant about a "Time Infrastructure"
<jbenet>
please don't use autotools oh god
<mmuller_>
PayasR: not so much, the toolchain works on Widnows.
<jbenet>
a makefile should be all you need. if you really need to get more intense. try out gyp or bazel
<mmuller_>
hehe, another satisfied customer :-)
<jbenet>
or npm.
<jbenet>
i think autotools left me with trauma.
<mmuller_>
you kinda need autotools for portability
<mmuller_>
yeah, whenever I have to deal with it I have to get over a deep sense of dread
<jbenet>
mmuller_ try out npm + gyp
<jbenet>
mmuller_ you won't look back.
<mmuller_>
jbenet: ok, I will
anshukla has joined #ipfs
<mmuller_>
Does gyp let you check for the presence of development libraries?
Wallacoloo has quit [Ping timeout: 246 seconds]
<PayasR>
mingw version of the toolchain works on windows.. yes.
<Blame>
and essentially IPFS already does all that stuff. without a whole lot of work
<jbenet>
Blame yeah. the last part needed is IPNS to work correctly, and as cjb (the author, who is a friend and who is in this channel) points out, the need for nice usernames (so some naming consensus).
<jbenet>
i'd be ok bootstrapping with email, pgp, keybase, github, twitter, or something else like that.
* cjb
waves
<cjb>
Yeah, it sounds like figuring out if GitTorrent can live on top of ipfs would be a neat next project for me. Gotta work out how to handle the git pack negotiation stuff.
<jbenet>
Blame: a draft of the record spec is out -- your feedback would be nice. i boiled it down a lot. this is like iteration 6.
<jbenet>
it defines records -- which themselves are merkledag objects!! -- that are "validity-scheme" agnostic, so can use whatever PKI and TI you have.
<whyrusleeping>
jbenet: you still didnt go over relative validity
<whyrusleeping>
which was mostly what i was waiting for :(
zen|merge has joined #ipfs
zen|merge has joined #ipfs
vijayee__ has joined #ipfs
Wallacoloo has quit [Quit: Leaving.]
notduncansmith has joined #ipfs
notduncansmith has quit [Read error: Connection reset by peer]
<jbenet>
whyrusleeping: ah yeah. i had, sec.
<jbenet>
whyrusleeping: so-- the "validity scheme" should order them?
hellertime has quit [Quit: Leaving.]
<jbenet>
whyrusleeping: ordering will be dependent on the TI anyway.
<whyrusleeping>
TI?
<whyrusleeping>
time infrastructure
<whyrusleeping>
got it
<wking>
jbenet: your "datetime range" and TTL approaches don't neccessarily help with ordering (e.g. you could have overlapping time ranges)
<whyrusleeping>
well, i would select the one with the range that will be valid the longest
<Blame>
The more we dig into this sort of spec the more I think it would be better suited to a domain-specific language like StackStream. Rather than "A bunch of meta info": bytecode that pulls a key from the local keystore and then decrypts the included binary blog somehow seems more generalized and sensical.
<Blame>
essentially replace merkle-dag and everything south of it.
<wking>
whyrusleeping: that works, but should probably be encoded in the spec. What do you do for TTLs?
<Blame>
I've been meaning to add "what minimum features does StackStream" need to be useful to ipfs to your todo repo.
<Blame>
s/blog/blob
<Blame>
essentially rip off a chunk of IPFS generalize it again.
<jbenet>
Blame what spec? the records?
<jbenet>
Blame: or the merkledag itself?
<Blame>
the merkledag itself.
<jbenet>
Blame: the merkledag is its own spec too, it's in that PR
<jbenet>
Blame: the beginnings of it anyway
<jbenet>
Blame: all of these are small WIPs.
<jbenet>
Blame: i've been very strict about keeping it very very small and understandable without delving too deeply.
<jbenet>
whyrusleeping, wking yeah agreed.
<Blame>
the best solution to time sycn is: just dont.
<vijayee__>
I believe I saw and actual repo for how they actual tutorials are to be structured. Does anyone know where it is?
<Blame>
put down a "valid untill" epoc time block and leave it at that. Then it is the signer's responsibility to know what time it is in order to have an accurate idea of how long the authorization will last and the client's responsibility to known when now is to prevent using outdated authorizations. "realitive" timeouts liek those used in DNS rely on trusting DNS
<Blame>
servers to stop relaying them after expiration.
<jbenet>
keychain is just a note, almost ignore it
<jbenet>
just note that keys + signatures are ipfs objects too
<whyrusleeping>
jbenet: i can look into that more. but last time i did i came to the conclusion that its most likely something to do with the environment on the travis VMs
<Blame>
jbenet: how do you transform the "keychain" merkle tree? say to publish a new name:key pair.
<jbenet>
whyrusleeping: are the tests giving us a lot of important info right now? or superfluous? maybe we can disable them if the latter until we can be sure they succeed?
<jbenet>
whyrusleeping: the other port stuff no longer clashes, i think there's something wrong with how iptb is assigning ports to things?
<whyrusleeping>
jbenet: well, iptb uses the different ports per test, but its the same ports across runs
<jbenet>
Blame: you add a node. it doesn't imply consensus. you'd resolve objects in this tree from people you trust (keyserver, WoT model). and if you want consensus (global trust) you use add roots to a well known tree ("blockchain")
<jbenet>
use->just
<Blame>
ok this makes a lot more sense. So it turns in trust-peer gossip? WOT has a fundamental issue :/ it is incentivised to re-centralize over time. Any way to stop that?
<wking>
jbenet: what's wrong with the protobuf schema format? You could even drop the field IDs if you felt they were too distracting, but I wouldn't bother.
<Blame>
you don't have to use the entire language, but the example at the bottom of stack stream shows a way to do byte-minimal merkledag blocks.
<vijayee__>
I'm seeing new activity around tour of ipfs. Is this created already?
<Blame>
right now, the length identifier is still more efficient than an ascii compatible encoding and using delimiters to indicate length (like json does)
<jbenet>
Blame: that's not a friendly format for our usage i think. i like the idea of using a language to describe it, but a serialized format should be optimized for fast traversal of links
<jbenet>
vijayee__ no, we should put it in another independent repo
<Blame>
So BFS over DFS?
<jbenet>
vijayee__ rht fixed some bugs, but i think we should use node workshopper or something like it in go
<jbenet>
Blame: no i just mean fast traversal from an implementation standpoint. byte offsets, memcpys, packets, etc.
<jbenet>
Blame: stuff you can optimize with fpgas etc.
<jbenet>
Blame: stackstream may be ok for this-- but im not sure of that yet. or of its ease of use for other thigns
<jbenet>
Blame i already have a hard enough time convincing people like ogd mafintosh and domanic that adopting a common format is good for all of us. asking them to use a bytecode language would probably make them kill me.
<vijayee__>
jbenet: ok well then I'll keep on the path I'm on
<jbenet>
Blame: i do appreciate the fundamental, computational generality though. wonder if it's possible to find a middle ground, define a language from something like protobuf or msgpack
notduncansmith has joined #ipfs
<jbenet>
(with probably a small header or something)
notduncansmith has quit [Read error: Connection reset by peer]
<jbenet>
Blame: so serializers can read the header manually, and use some serialization library straight up, but a vm could be super smart about it
<krl>
jbenet: need cr the most @ ipfs-api and ipfsd-ctl
<kyledrake>
Good afternoon!
<Tv`>
my current favorite bytecode is x86
<jbenet>
llvm is another good option :)
<jbenet>
Tv` btw most x86 archs now actually use a RISC core, but they dont call it that cause CISC won the war.
<jbenet>
kyledrake: hello! o/
<jbenet>
krl: links to PRs or commits to CR pls?
<Tv`>
fwiw the old argument, from the people who designed it, is that LLVM IR is not portable and was not meant to be a bytecode
www has joined #ipfs
www1 has quit [Ping timeout: 255 seconds]
<Tv`>
jbenet: there's a reason i called x86 a bytecode..
<jbenet>
Tv` ah hahaha
<jbenet>
Tv` also the pnacl folks made that LLVM IR work pretty portably.
<Tv`>
sure, though i'd call PNaCl more LLVM IR-inspired than LLVM IR
<jbenet>
you're ignoring it. it shouldn't, but it might under memory pressure.
inconshr_ has joined #ipfs
Confis has joined #ipfs
<jbenet>
whyrusleeping what sync problems are there with `msg` ?
<jbenet>
i see a lock being taken for s but not sure on msg
inconshreveable has quit [Ping timeout: 258 seconds]
kbala_ has joined #ipfs
mildred has quit [Quit: Leaving.]
notduncansmith has joined #ipfs
notduncansmith has quit [Read error: Connection reset by peer]
tilgovi has quit [Remote host closed the connection]
chriscool has quit [Ping timeout: 252 seconds]
ei-slackbot-ipfs has quit [Remote host closed the connection]
ei-slackbot-ipfs has joined #ipfs
<cryptix>
re
kyledrake has quit [Ping timeout: 245 seconds]
pfraze has joined #ipfs
void has joined #ipfs
<jbenet>
whyrusleeping did you see the commit i pushed re ordering?
pfraze has quit [Remote host closed the connection]
<jbenet>
whyrusleeping: does that satisfy the concerns? / make sense?
<jbenet>
whyrusleeping: what else is needed on record spec?
pfraze has joined #ipfs
notduncansmith has joined #ipfs
<whyrusleeping>
jbenet: yeah, i saw the ordering commit
notduncansmith has quit [Read error: Connection reset by peer]
<whyrusleeping>
and there arent any errors returned from the buf.Write
<whyrusleeping>
sprint checkin: cleaning up and finalizing open PRs, debugging network performance issues
<cryptix>
jbenet: re listen->print: the order of the two prints changed which makes the tests fail m(
<cryptix>
sprint checkin: call me leeroy godeps jenkins
<jbenet>
sprint checkin: pushed [record, merkledag] spec, some CR, behind on major CR items (S3 and pinning and migration)
<jbenet>
cryptix: thanks for all that
<jbenet>
whyrusleeping maybe we should have a sprint bot
<whyrusleeping>
jbenet: i can do that
<jbenet>
have it ask us every day.
<jbenet>
and maybe it can ping people.
<jbenet>
registered on the sprint that week or something.
<jbenet>
one option is to checkin on github
<jbenet>
instead of irc
<jbenet>
(bot could post to that too)
<cryptix>
jbenet: very welcome. will read the new specs later. i will put the api setup before gateway - we need to reorder a lot of sharness tests otherwise, i fear
<cryptix>
you guys should checkout ssb btw
www1 has joined #ipfs
www has quit [Ping timeout: 256 seconds]
void has quit [Quit: ChatZilla 0.9.91.1 [Firefox 38.0/20150511103818]]
notduncansmith has joined #ipfs
notduncansmith has quit [Read error: Connection reset by peer]
pfraze has quit [Remote host closed the connection]
<jbenet>
yeah i've seen it -- i was one of the first users :) <3
inconshr_ has quit [Ping timeout: 256 seconds]
<jbenet>
whyrusleeping: btw the keystore can be objects in the repo itself, like pins. we'd need explicit `ipfs key import` and `ipfs key export`, but then we can have the keys be part of the repo itself. we can encrypt secret keys before storing the objects.
lgierth_ has joined #ipfs
lgierth_ has quit [Client Quit]
lgierth has quit [Ping timeout: 258 seconds]
<cryptix>
whyrusleeping: re 1304 - i'm not sure how you buffer up anything with that change. you just copy it once more into the buffer before it goes into the writer below
<ipfsbot>
[go-ipfs] cryptix force-pushed maint/fuse from 8ae4283 to e9074be: http://git.io/vkuIz
<cryptix>
jbenet: yea - i feel like they degraded a little again
inconshreveable has joined #ipfs
lgierth has joined #ipfs
<jbenet>
its all the random failures
<jbenet>
its better but not fully there yet
<Tv`>
current status: "git rebase baster"
notduncansmith has joined #ipfs
notduncansmith has quit [Read error: Connection reset by peer]
<cryptix>
jbenet: do we really need 'ipfs daemon --init' i feel like 'ipfs init && ipfs config ... && ipfs daemon' is easier than adding even more flags to the daemon cmd
lgierth has quit [Quit: Ex-Chat]
<jbenet>
cryptix: what if you have to run the _exact__ command twice?
<ipfsbot>
[go-ipfs] jbenet deleted maint/fuse at e9074be: http://git.io/vkzTN
notduncansmith has joined #ipfs
notduncansmith has quit [Read error: Connection reset by peer]
<cryptix>
jbenet: this feels like a trick question
<jbenet>
cryptix: no, it isn't. people need something to put in their scripts that will run every time.
<jbenet>
cryptix: so "ipfs daemon --init < initial-config" is a real use case.
<jbenet>
you want it to init and set the config only the very first time.