seg_violation has quit [Remote host closed the connection]
rhalff has joined #ipfs
cryptix has joined #ipfs
kalkin-_ has joined #ipfs
<kalkin-_>
hi
kalkin-_ is now known as kalkin-
<kalkin->
So I'm looking at IPNS. Part of it's name is Naming System. My assumption when hearing the name is that it is DNS like. But from reading the example I don't see any benefits in using IPNS.
<kalkin->
How would I use ipns?
<kalkin->
it's not like i can do ipnfs name resolve my-simple-name and it will return me the hash
<ylp>
it make an indirection between a static hash and the hash of a content
<kalkin->
As far as i understand it's just a way to have a reference to the up to date object you are storing in ipfs
<ylp>
yes
<kalkin->
reminds me of Freenets USK updatable subspace key
<kalkin->
ylp: ic. thanks for confirmation
<kalkin->
you guys probably want to rename ipns to something different, even if the name makes sense to the developers, other people might get confused.
<ylp>
and as I understood you can only had one per node...
<kalkin->
generally IPFS looks to me from very high above like Freenet without anonymity and Java. (Yes i know Freenet uses partly different technology)
<Kubuxu>
kalkin-: IPNS allows for use of DNS too.
<Kubuxu>
you can create DNS record: `_dnslink.example.com TXT dnslink=/ipfs/Qm..AAA` and it will use that
<cow_2001>
`(Hit ctrl-c again to force-shutdown the daemon.)` <- to what length will my horns grow if i hit ctrl-c again?
rhalff has joined #ipfs
rardiol has joined #ipfs
corvinux has joined #ipfs
A124 has quit [Quit: '']
A124 has joined #ipfs
xelra has quit [Ping timeout: 260 seconds]
cemerick has joined #ipfs
stoogenmeyer has joined #ipfs
xelra has joined #ipfs
<chungy>
Freenet and IPFS do have some similarities with regard to content distribution, but IPFS totally ignores the anonymity aspects in favor of speed -- ideally, IPFS should be able to transfer data as fast as the available hardware (disks, networks)
stoogenmeyer has quit [Ping timeout: 240 seconds]
j12t has joined #ipfs
jan__ has joined #ipfs
j12t has quit [Ping timeout: 272 seconds]
Kane` has quit [Quit: Leaving]
corvinux has quit [Ping timeout: 244 seconds]
keorn has quit [Remote host closed the connection]
unforgiven512_ has quit [Ping timeout: 240 seconds]
<lgierth>
ipfs isn't ignorant of anonymity -- it just leaves it to the network layer (libp2p)
<lgierth>
there's a patch for tor/libp2p integration flying around somewhere
<kalkin->
lgierth: meh patches doesn't count. :)
<lgierth>
fair enough
chriscool1 has joined #ipfs
chriscool1 has quit [Client Quit]
chriscool1 has joined #ipfs
kerozene has quit [Ping timeout: 260 seconds]
chriscool has quit [Quit: Leaving.]
rardiol has quit [Ping timeout: 244 seconds]
kerozene has joined #ipfs
<chungy>
lgierth: ipsec should be an option too...
chriscool1 has quit [Quit: Leaving.]
chriscool has joined #ipfs
pyron has joined #ipfs
rhalff has quit [Ping timeout: 260 seconds]
zorglub27 has quit [Quit: zorglub27]
rhalff has joined #ipfs
mildred has joined #ipfs
cemerick has quit [Ping timeout: 272 seconds]
computerfreak has joined #ipfs
jan__ has quit [Quit: Page closed]
pyron has quit [Quit: leaving]
nonaTure has joined #ipfs
ygrek has joined #ipfs
nonaTure has quit [Ping timeout: 252 seconds]
cryptix has quit [Ping timeout: 244 seconds]
computerfreak has quit [Remote host closed the connection]
computerfreak has joined #ipfs
rhalff has quit [Ping timeout: 252 seconds]
kanzure_ is now known as kanzure
ygrek has quit [Ping timeout: 240 seconds]
rgrinberg has joined #ipfs
cemerick has joined #ipfs
apiarian has quit [Ping timeout: 244 seconds]
computerfreak has quit [Quit: Leaving.]
apiarian has joined #ipfs
<Kubuxu>
chungy: ipsec is not doing what it is supposed to be doing unfortunately
stoogenmeyer has joined #ipfs
PrinceOfPeeves has joined #ipfs
stoogenmeyer has quit [Read error: Connection reset by peer]
stoogenmeyer has joined #ipfs
dmr has quit [Ping timeout: 250 seconds]
special has quit [Ping timeout: 258 seconds]
j12t has joined #ipfs
ruby32 has quit [Ping timeout: 276 seconds]
special has joined #ipfs
special is now known as Guest37019
j12t has quit [Ping timeout: 260 seconds]
kalkin- has quit [Ping timeout: 276 seconds]
ruby32 has joined #ipfs
kerozene has quit [Quit: WeeChat 1.4-dev]
rhalff has joined #ipfs
zz_r04r is now known as r04r
cryptix has joined #ipfs
xelra has quit [Ping timeout: 260 seconds]
rardiol has joined #ipfs
passstab has quit [Ping timeout: 252 seconds]
xelra has joined #ipfs
dguttman has joined #ipfs
cemerick has quit [Ping timeout: 240 seconds]
_rht has joined #ipfs
reit has quit [Ping timeout: 276 seconds]
cketti has joined #ipfs
cketti has quit [Changing host]
cketti has joined #ipfs
computerfreak has joined #ipfs
ruby32 has quit [Ping timeout: 260 seconds]
cemerick has joined #ipfs
dguttman has quit [Quit: dguttman]
reit has joined #ipfs
ashark has joined #ipfs
HostFat has joined #ipfs
cryptix has quit [Quit: leaving]
cemerick has quit [Ping timeout: 252 seconds]
Guest19517 has joined #ipfs
fredthomsen has quit [Ping timeout: 252 seconds]
xelra has quit [Ping timeout: 260 seconds]
xelra has joined #ipfs
ruby32 has joined #ipfs
stoogenmeyer has quit [Remote host closed the connection]
<ipfsbot>
[go-ipfs] Kubuxu pushed 1 new commit to feature/update-authors: https://git.io/voixK
<ipfsbot>
go-ipfs/feature/update-authors d54c3fa Jakub Sztandera: Remove one of my clones and one of Wwhy's...
<ipfsbot>
[go-ipfs] Kubuxu force-pushed feature/update-authors from d54c3fa to 251918d: https://git.io/vogiq
<ipfsbot>
go-ipfs/feature/update-authors 251918d Jakub Sztandera: Remove one of my clones and one of Why's...
<Kubuxu>
richardlitt: wait a sec on go-ipfs
<Kubuxu>
richardlitt: now it is LGTM
<ipfsbot>
[go-ipfs] Kubuxu force-pushed feature/update-authors from 251918d to 92cd0f9: https://git.io/vogiq
<ipfsbot>
go-ipfs/feature/update-authors 92cd0f9 Jakub Sztandera: Remove one more clone of mine...
<richardlitt>
Thanks Kuba
slothbag1 has quit [Quit: Leaving.]
Boomerang has joined #ipfs
Boomerang has quit [Client Quit]
Boomerang has joined #ipfs
dguttman has joined #ipfs
stoogenmeyer has quit [Remote host closed the connection]
Arw_ is now known as Arwalk
dguttman has quit [Quit: dguttman]
Peer3Peer has quit [Ping timeout: 260 seconds]
stoogenmeyer has joined #ipfs
dguttman has joined #ipfs
j12t has joined #ipfs
j12t has quit [Ping timeout: 252 seconds]
kerozene has joined #ipfs
Boomerang has quit [Quit: Leaving]
cow_2001 has quit [Quit: cow_2001]
tlevine_ has joined #ipfs
<ipfsbot>
[go-ipfs] RichardLitt force-pushed feature/github-issues-guide from b21104e to d8109b6: https://git.io/voPt5
<ipfsbot>
go-ipfs/feature/github-issues-guide d8109b6 Richard Littauer: Added a How To File GitHub Issues guide...
<ipfsbot>
[go-ipfs] RichardLitt opened pull request #2889: Added a How To File GitHub Issues guide (master...feature/github-issues-guide) https://git.io/voPtb
<whyrusleeping>
bret: yeah!!! try clicking the stars
<bret>
whyrusleeping: nodda
<bret>
whyrusleeping: but they are connecting now
<whyrusleeping>
hrm... i'm not quite sure how the click event gets handled
<whyrusleeping>
but its supposed to shoot out lines from stars you click
<whyrusleeping>
i think it takes a long click
<whyrusleeping>
oh well
<bret>
let me try chrome
<bret>
¯\_(ツ)_/¯
<bret>
still looks great
rhalff has quit [Remote host closed the connection]
<whyrusleeping>
thanks :)
<whyrusleeping>
i'm mostly happy you can get an idea of what ipfs is from reading the website
<lgierth>
that star animation is still eating my cpu
* lgierth
inclined to deactivate javascript
Pentb has joined #ipfs
s_kunk has joined #ipfs
Pent has quit [Ping timeout: 244 seconds]
gmcquillan has joined #ipfs
<Kubuxu>
whyrusleeping: will the same message appear with `ipfs get` and `ipfs object get`? I think yes.
<whyrusleeping>
Kubuxu: yeap
<whyrusleeping>
lgierth: doesnt your chromebooks processor have the 'super star drawing' cpu extensions?
<Kubuxu>
whyrusleeping: seems good, I can't think off anything better but I am bad in that.
<whyrusleeping>
Kubuxu: cool cool
<whyrusleeping>
also, 2814 or 2852?
<whyrusleeping>
theyre functionally equivalent at this point
<Kubuxu>
yeah, I don't know what to think about it.
<Kubuxu>
(We can filter Get misses with bloom filter too :D)
<ipfsbot>
[go-ipfs] whyrusleeping pushed 1 new commit to feat/not-object-error: https://git.io/voP1G
<ipfsbot>
go-ipfs/feat/not-object-error 306f766 Jeromy: add a little bit more verbosity to the error...
emma has joined #ipfs
<whyrusleeping>
Kubuxu: can we though? I think that would require us to always have every key in our repo inserted in the bloom filter
<whyrusleeping>
which now that i think about it is probably always the case
<Kubuxu>
We have to have every key in the repo inserted into the bloom.
<whyrusleeping>
ouch
<Kubuxu>
Otherwise bloom is useless.
<whyrusleeping>
right
<Kubuxu>
now I do walk over keys on the startup, I think it is best way to stay in sync with the repo
<whyrusleeping>
thats gonna make startup take ages
<Kubuxu>
It does it in separate goroutine.
<whyrusleeping>
hrmmm
<whyrusleeping>
and just doesnt have a bloom filter set until that finishes?
<Kubuxu>
and until it is finished bloom is bypassed
<whyrusleeping>
okay
<whyrusleeping>
it might make sense to persist the bloom filter on disk in this case
<whyrusleeping>
but that just leads to problems of being out of date
<Kubuxu>
There is no way of knowing if there are additional blocks in the blockstore then.
<whyrusleeping>
hrm...
<whyrusleeping>
the issue i see here is that this bloom filter is now adding a ton of complexity...
<Kubuxu>
It might be worth implementing, on go-datastore/flatfs, key index.
galois_dmz has joined #ipfs
<whyrusleeping>
yeah, that might be a good place for it
Boomerang has quit [Quit: Leaving]
galois___ has joined #ipfs
<Kubuxu>
Now the question: should I put normal cache before and after bloom? I've checked they have similar access times.
<Kubuxu>
~1 micro second
<whyrusleeping>
well, that depends on how much time a cache failure costs
<whyrusleeping>
and the frequency of each
galois_d_ has quit [Ping timeout: 272 seconds]
<whyrusleeping>
or alternatively, how much time does a cache hit save us
<whyrusleeping>
i think the lru cache saves us much more time on a cache hit than a 'hit' on the bloom filter
galois_dmz has quit [Ping timeout: 260 seconds]
<Kubuxu>
Of all requests we have 98% of return false, which will be hits on bloom (with <1% false positive).
<Kubuxu>
So placing bloom first eliminates ~90% of requests, so bloom first IMO.
<whyrusleeping>
fair
locusf has quit [Read error: Connection reset by peer]
<ipfsbot>
[go-ipfs] kevina opened pull request #2890: [WIP] Implement Basic LRU Cache for merkledag nodes. (master...dag-cache) https://git.io/voPys
<Kubuxu>
I don't know how many of them are repeated requests but I don't think 90%.
<Kubuxu>
whyrusleeping: should I make it LRU or something more fancy (ARC, 2Q)?
arpu_ has quit [Ping timeout: 258 seconds]
xelra has quit [Ping timeout: 260 seconds]
xelra has joined #ipfs
kevina has joined #ipfs
Encrypt has quit [Quit: Quitte]
Encrypt has joined #ipfs
arpu_ has joined #ipfs
<Kubuxu>
Looks I will obsolete write_cached
<ipfsbot>
[go-ipfs] whyrusleeping created feat/ulimit-raise-tests (+1 new commit): https://git.io/voPHs
<ipfsbot>
go-ipfs/feat/ulimit-raise-tests 3d5e015 Jeromy: add a test for auto-raising ulimit...
<whyrusleeping>
Kubuxu: yeah, i'm thinking that we should just expand the usefulness of the writecache
<whyrusleeping>
lets keep it as an LRU for now
<whyrusleeping>
we already have an LRU lib we're importing and its working well for us, if a better caching structure is called for later we can always change it
<Kubuxu>
That lru lib contains ARC
<Kubuxu>
ARC is LRU that has one timers rush prevention.
<Kubuxu>
s/rush/burst
<whyrusleeping>
Kubuxu: ah, i see
<whyrusleeping>
I would be fine trying out the arc cache then
<ipfsbot>
[go-ipfs] whyrusleeping opened pull request #2891: add a test for auto-raising ulimit (master...feat/ulimit-raise-tests) https://git.io/voPQb
<michaeldgagnon>
reading about ipfs for the first time and trying to fit it in my head... am I correct in my interpretation that, if I have 5gb asset, that is broken up into a bunch of blocks. If you pull that, you get the metadata pointing at all the blocks and you get all the blocks. If I change 1 byte in that file, it's a new hash. If you pull that new file, you get the new metadata which has only 1 different block. You pull a single block fo
<michaeldgagnon>
Is that a correct reading?
Encrypt has quit [Quit: Quitte]
emma has quit [Client Quit]
emma has joined #ipfs
fleeky has quit [Ping timeout: 250 seconds]
emma has quit [Client Quit]
emma has joined #ipfs
cemerick has joined #ipfs
<Kubuxu>
michaeldgagnon: almost, data is stored in binary tree where branches are hashes with final leafs being your binary data, so you will have to get whole branch.
galois___ has quit [Remote host closed the connection]
galois_dmz has joined #ipfs
stoogenmeyer has quit [Remote host closed the connection]
<A124>
For when is planned (if is) the next point (4.3 or 4.2.x) release?
<Kubuxu>
there is no further x
<Kubuxu>
it is 0.4.2 now
<Kubuxu>
next is 0.4.3
stoogenmeyer has joined #ipfs
<whyrusleeping>
next release is planned second week of july
<whyrusleeping>
michaeldgagnon: that is roughly correct, as Kubuxu said, you will have to fetch all the tree nodes that have your changed block as a descendant
<whyrusleeping>
which is generally only one or two blocks, unless your files are massive
<whyrusleeping>
then it could be three
<Kubuxu>
Heh :p
<whyrusleeping>
(yay exponentiation)
rendar has quit [Ping timeout: 244 seconds]
<A124>
Thanks.
devbug has joined #ipfs
cemerick has quit [Ping timeout: 276 seconds]
<ipfsbot>
[go-ipfs] RichardLitt force-pushed feature/github-issues-guide from d8109b6 to d785b4e: https://git.io/voPt5
<ipfsbot>
go-ipfs/feature/github-issues-guide d785b4e Richard Littauer: Added a How To File GitHub Issues guide...
<ipfsbot>
[go-ipfs] RichardLitt force-pushed feature/github-issues-guide from d785b4e to e253190: https://git.io/voPt5
<ipfsbot>
go-ipfs/feature/github-issues-guide e253190 Richard Littauer: Added a How To File GitHub Issues guide...
zorglub27 has joined #ipfs
<whyrusleeping>
Kubuxu: i'm thinking that we should focus on the caching improvements for the blockstore stuff before worry about the bloom filter
<whyrusleeping>
since the bloom filter will be more complicated and take more time to be tested thoroughly, and we want some of these improvements into the 0.4.3 release
<Kubuxu>
I am just finishing both, better caching and bloom.
<whyrusleeping>
okay
<ipfsbot>
[go-ipfs] RichardLitt force-pushed feature/github-issues-guide from e253190 to 715b614: https://git.io/voPt5
<ipfsbot>
go-ipfs/feature/github-issues-guide 715b614 Richard Littauer: Added a How To File GitHub Issues guide...
Pentb has quit [Quit: Leaving]
* Kubuxu
right now: "xflux, NO, I have a work to do"
<Kubuxu>
"pkill xflux"
Pent has joined #ipfs
rendar has joined #ipfs
<ipfsbot>
[go-ipfs] whyrusleeping closed pull request #2889: Added a How To File GitHub Issues guide (master...feature/github-issues-guide) https://git.io/voPtb
<ipfsbot>
[go-ipfs] whyrusleeping deleted feat/not-object-error at 306f766: https://git.io/voPhQ
<A124>
whyrusleeping Were you working on the ... context switching problem, or it is jut a mix of circumstances?
dguttman has quit [Quit: dguttman]
<A124>
I have not tested in regular mode, but when in lowmemory and no reuse mode on about a week old code, the context noise was at ok level, except massive (50k) spikes every two minutes.
<whyrusleeping>
A124: i have improved things on that end, yes
<whyrusleeping>
youre running latest master?
<whyrusleeping>
"week old"
<whyrusleeping>
Kubuxu: i love that unix gives us the ability to kill things
<Kubuxu>
Yup, I love xflux, when I am not working.
<Kubuxu>
It really puts me to sleep.
<A124>
whyrusleeping Well multiple places. It's jbenet's docker.
<A124>
On the one I develop I do not have time for steady provable measurements.
<ipfsbot>
[go-ipfs] Kubuxu pushed 1 new commit to feature/bloom-cache: https://git.io/voXev
<ipfsbot>
go-ipfs/feature/bloom-cache 4765754 Jakub Sztandera: Replace write_cache with bloom_cache...
<A124>
whyrusleeping But well done and thank you, for everyone. It was one of the things really needed. will have to check the spikes, when they are gone, I would consider that runnable for more serious stuff. memory leakage can be solved by occasional restarts and lowmem mode, unless it is that... ipfs add problem.
<A124>
Looking forward using ipfs as a base for other applications, including those merely using ipfs to find a peer by ID and make a tunnel in between.
<Kubuxu>
whyrusleeping: 4 builds in teamcity queue
<Kubuxu>
`ipfs repo stat` just took 30min on my server
<ipfsbot>
[go-ipfs] Kubuxu pushed 1 new commit to feature/bloom-cache: https://git.io/voXBw
<ipfsbot>
go-ipfs/feature/bloom-cache c226019 Jakub Sztandera: Skip test too...
Boomerang has quit [Quit: Leaving]
Stebalien has quit [Remote host closed the connection]
zorglub27 has joined #ipfs
_rht has quit [Quit: Connection closed for inactivity]
<Kubuxu>
whyrusleeping: do we want to fix#2601 before 0.4.3?
<Kubuxu>
yes we want, never mind
zorglub27 has quit [Client Quit]
<whyrusleeping>
Kubuxu: affirmative
<whyrusleeping>
i'm working on the fix right now
<whyrusleeping>
the fix is easy
<whyrusleeping>
its the migration thats annoying
<Kubuxu>
but we need to deploy repo converter
galois_dmz has joined #ipfs
<Kubuxu>
also: any decisions on the stdin StringArg handing?
galois___ has joined #ipfs
ligi has joined #ipfs
galois_d_ has quit [Ping timeout: 252 seconds]
<ligi>
Is there a way to speed up ipfs name/publish?
j12t has joined #ipfs
<Kubuxu>
ligi: no until pubsub comes out, ipfs name publish takes a while because it connects with +20 nodes close to you in key space.
galois_dmz has quit [Ping timeout: 258 seconds]
<ligi>
Kubuxu: thanks
<Kubuxu>
and requests acknowledge from those nodes that they will keep record, IIRC.
<lgierth>
hey ligi o/
<ligi>
hey lgierth
<lgierth>
it's my second day as a c-base member :D
<ligi>
lgierth: great!
<lgierth>
i'm larse
<kevina>
Kubuxu: I basically agree with you on StringArg handing. A few utilities allow reading of arguments from a file, but there are special options for that.
<lgierth>
lars and larsg seem to have been taken
<ligi>
and lgierth was also taken?
<lgierth>
it's so hard to spell
<lgierth>
but larse is cool actually
<ligi>
that's right - but now there is a mapping needed ;-)
<lgierth>
:)
<lgierth>
police in your street too?
<ligi>
lgierth: I have no window to the outside
<ligi>
and I do not want to go out at the moment
<ligi>
but I suspect so
<lgierth>
:)
<ligi>
rigaer
<lgierth>
ooh really
<ligi>
yea
<lgierth>
if you need place to crash or so -- we have plenty of space
<ligi>
300 police man this morning
cow_2001 has quit [Quit: cow_2001]
<ligi>
no - I do not live there ..
<ligi>
but thanks
<lgierth>
aah ok
<lgierth>
yeah i'm pretty familiar with the whole situation :D
<ligi>
(y)
<Kubuxu>
lgierth: what is going on?
<ligi>
they where cutting the internet cable there - not cool - happy to not live there
<ligi>
Kubuxu: politrixx
<lgierth>
Kubuxu: emery's neighbours got evicted today -- same house as in february
Guest37019 is now known as special
<lgierth>
and judging from the activity at the police station close to my place, it's gonna be a long night
<whyrusleeping>
apiarian: definitely some weird stuff going on there... could you post that to the issue so i don't lose the link?
<apiarian>
whyrusleeping: will do
<whyrusleeping>
Kubuxu: could you do a small writeup on what the `EnableStdin()` removal would change?
<whyrusleeping>
Its going to be a fairly major change to the api if we go that way
<whyrusleeping>
(or at least, a major change to the cli interface)
<apiarian>
whyrusleeping: separate question: can i submit a patch to https://github.com/ipfs/go-ipfs-api/blob/master/ipns.go#L27 to not send id if it is an empty string to name/resolve so that we can use the "resolve-self" functionality?
<kpcyrd>
dansup: also, if you add a setting to the config to POST the returning ipfs path to an external $url, people can use this to connect a pin-service (if somebody is building something like this)
stoogenmeyer has joined #ipfs
stoogenm_ has joined #ipfs
matoro has joined #ipfs
stoogenm_ has quit [Remote host closed the connection]