<lgierth>
i was hoping for more technical detail ;)
matoro has quit [Ping timeout: 260 seconds]
matoro has joined #ipfs
ashark has quit [Ping timeout: 276 seconds]
gigq_ has quit [Ping timeout: 276 seconds]
gigq has joined #ipfs
<zero-ghost>
articles 1 and 2 give a little more tech detail but his series isnt the best for highly technical details
dignifiedquire has quit [Quit: Connection closed for inactivity]
cketti has quit [Quit: Leaving]
wallacoloo has joined #ipfs
dmr has joined #ipfs
Jupely has left #ipfs [#ipfs]
wallacoloo has quit [Quit: wallacoloo]
matoro has quit [Ping timeout: 246 seconds]
wking has quit [Ping timeout: 260 seconds]
wking has joined #ipfs
matoro has joined #ipfs
PrinceOfPeeves has quit [Quit: Leaving]
ligi_ has joined #ipfs
ghtdak has quit [Ping timeout: 276 seconds]
ghtdak has joined #ipfs
ligi has quit [Ping timeout: 250 seconds]
M-jackal has joined #ipfs
Pent has joined #ipfs
taaem has quit [Read error: Connection reset by peer]
MadcapJake has quit [Quit: MadcapJake]
jrusso has joined #ipfs
taaem has joined #ipfs
jrusso is now known as MadcapJake
slothbag has quit [Quit: Leaving.]
dmr has quit [Ping timeout: 250 seconds]
gigq has quit [Ping timeout: 260 seconds]
gigq has joined #ipfs
taaem has quit [Ping timeout: 260 seconds]
herzmeister has quit [Ping timeout: 244 seconds]
redfish has quit [Ping timeout: 264 seconds]
wallacoloo has joined #ipfs
* A124
Today I did bump into the encoding of IDs (Qm...), now cannot bump into it, if someone could... thanks
mpi has joined #ipfs
gigq has quit [Ping timeout: 250 seconds]
gigq has joined #ipfs
devbug has quit [Ping timeout: 260 seconds]
disgusting_wall has quit [Quit: Connection closed for inactivity]
mpi_ has joined #ipfs
mpi has quit [Ping timeout: 250 seconds]
<Kubuxu>
apiarian: as when you publish, you push the record to 20 nodes, and when someone wants to find a record, he checks ~10, there is really good chance that it will be the newest/
dignifiedquire has joined #ipfs
mpi_ has quit [Remote host closed the connection]
mpi has joined #ipfs
mpi has quit [Remote host closed the connection]
luigiplr has quit [Remote host closed the connection]
reit has quit [Ping timeout: 276 seconds]
sahib has joined #ipfs
Tv` has quit [Quit: Connection closed for inactivity]
<deltab>
A124: base58 using the bitcoin alphabet
Senji has joined #ipfs
computerfreak has joined #ipfs
zz_r04r is now known as r04r
toxip has joined #ipfs
ligi has joined #ipfs
ligi_ has quit [Ping timeout: 264 seconds]
wallacoloo has quit [Quit: wallacoloo]
rendar has joined #ipfs
ralphtheninja has joined #ipfs
ygrek has joined #ipfs
zorglub27 has joined #ipfs
cemerick has joined #ipfs
rdewilde has joined #ipfs
reit has joined #ipfs
zorglub27 has quit [Ping timeout: 276 seconds]
rdewilde has quit [Ping timeout: 250 seconds]
herzmeister has joined #ipfs
tdmps has joined #ipfs
tdmps has left #ipfs [#ipfs]
computerfreak has quit [Quit: Leaving.]
Encrypt has joined #ipfs
computerfreak has joined #ipfs
computerfreak has quit [Client Quit]
apiarian has quit [Quit: zoom]
slothbag has joined #ipfs
computerfreak has joined #ipfs
jonnor has joined #ipfs
ralphtheninja has quit [Quit: leaving]
zorglub27 has joined #ipfs
Aeonwaves has quit [Ping timeout: 260 seconds]
Aeon has joined #ipfs
Aeon has quit [Changing host]
Aeon has joined #ipfs
<jonnor>
what is the current block size in the IPFS protocol? meaning where if bigger, it will be split into multiple
ligi_ has joined #ipfs
<ligi_>
hi - is there a way to query if the daemon is ready to use?
palkeo_ has joined #ipfs
M-wkennington is now known as wkennington-m
M-jackal is now known as cwopel
<Kubuxu>
jonnor: 256KB
<Kubuxu>
ligi_: depends what do you mean by ready to use
<Kubuxu>
ligi_: I poll `/api/v0/version`.
jonnor has quit [Ping timeout: 260 seconds]
apiarian has joined #ipfs
cketti has joined #ipfs
cemerick has quit [Ping timeout: 276 seconds]
<apiarian>
Kubuxu: what about my own node? does it always respond with the latest value of its own record?
<apiarian>
(to local api requests)
<Kubuxu>
It will also run the search, as you could have for example switched machines (or published the record via other means).
sahib has quit [Ping timeout: 272 seconds]
<apiarian>
Kubuxu: i can run multiple nodes with the same id?
<Kubuxu>
You could try, what you can do is run a node that start, publish and exit. There is even program for that,
victorbjelkholm has quit [Remote host closed the connection]
victorbjelkholm has joined #ipfs
matoro has quit [Ping timeout: 276 seconds]
matoro has joined #ipfs
Oatmeal has joined #ipfs
Aaron1011 has joined #ipfs
dguttman has quit [Quit: dguttman]
rhalff has quit [Ping timeout: 260 seconds]
slothbag has quit [Quit: Leaving.]
<apiarian>
why do ipns entries have a default lifetime of 24 hours? why have a default expiration time on them at all?
<Kubuxu>
To prevent someone separating you from the new version.
<apiarian>
how's that?
<Kubuxu>
Imagine scenario when you release software under ipns, there is one with a serious bug, if you were out of the network (or I was able to separate you from connecting to the rest of the network, quite easy, possible on WPA-PSK) I would be able to show you only the old, vulnerable record.
<apiarian>
ah
<Kubuxu>
IPRS (record system) which is to replace IPNS will allow for customization of this metric (validity metric), but for now it is 24h.
<apiarian>
so if i want to keep a record current, i need to periodically re-publish it
<Kubuxu>
It should be republished automagically by ipfs daemon.
<apiarian>
oh? i published something to QmNT744VjtRFpDYB25EVLx7ha1zAVDKsd3qFjxfQLjPEXq with a lifetime of 30 seconds and it is no longer resolvable
<Kubuxu>
Using ipns-pub or daemon?
<apiarian>
oh, no, using go-ipfs
MadcapJake has quit [Quit: MadcapJake]
<Kubuxu>
Hmm, the republisher might be set on constant time.
<Kubuxu>
Like 12h or something.
jrusso has joined #ipfs
jrusso is now known as MadcapJake
<Kubuxu>
Yup, 4h
<apiarian>
ah, so if i have my go-ipfs daemon running and i publish something with a lifetime > 4 hours it should theoretically keep republishing it indefinitely? (does that persist across daemon restarts?)
chriscool1 has joined #ipfs
<Kubuxu>
It should be republished, I don't know how about restarts unfortuantelly.
jonnor has joined #ipfs
herzmeister has quit [Quit: Leaving]
herzmeister has joined #ipfs
<apiarian>
Kubuxu: could you point me to the area of the code this all happens?
<M-leer10>
though I'm not even sure why they do that: how would anyone be convinced by some hit-and-run one-liner to do something in a foreign language
<M-leer10>
Kubuxu_ (IRC): That's cool
<kaoD>
can anyone TL;DR me how does IPFS compare to Telehash?
<kaoD>
AFAICT IPFS is similar to Telehash but oriented towards a hierarchical FS while Telehash is just a content-agnostic transport... did I get that right?
<Kubuxu>
kaoD: I don't know Telehash, but just by looking at it Telehash looks like just a transport protocol.
<Kubuxu>
IPFS aims at moving content around
Tristitia has joined #ipfs
<stoogenmeyer>
Kubuxu: so that copy of OSM is immutable, though, right?
<Kubuxu>
stoogenmeyer: yes, but we can use ipns to allow for updating it.
<kaoD>
I see thanks
<stoogenmeyer>
you mean ipns will resolve to a new hash with a newer version?
<stoogenmeyer>
but then the old content is still in existence?
<Kubuxu>
yup, but much of the old content will be included in new content.
<stoogenmeyer>
aww.. missed that crucial point
<Kubuxu>
And IPFS will be able to dedup it a lot (at least I think so).
<Kubuxu>
For example cdnjs (CDN package of js libs) gets deduped from 21GB to about 16GB in IPFS storage.
<stoogenmeyer>
huh, neat. how were you guys able to find that out?
<Kubuxu>
I cloned cjdns to my server, added to to IPFS and checked the size of internal IPFS storage..
<Kubuxu>
s/cjdns/cdnjs
<stoogenmeyer>
that implies that you were the first to add the newest cdnjs version to ipfs, though, right? otherwise your local repo would be empty?
ashark has joined #ipfs
<Kubuxu>
If you add something to IPFS, it is stored in 100% on your machine, IPFS starts to distribute content only when someone requests it.
<stoogenmeyer>
wait but what happens if someone already added that content previously? e.g. I add a text file with "hello world", which someone has already done before. Will I get the same hash that he did?
<Kubuxu>
Yes you will
<stoogenmeyer>
so going back to the previous point.. If someone already added the latest version of cdnjs, and someone else asked for it (how would they know what to ask for though?), then it would be already distributded across ipfs. so when you add it as well, why would there be a need for it to exist in your local repo? if you can fetch it from ipfs?
herzmeister has quit [Quit: Leaving]
herzmeister has joined #ipfs
freemint_ has joined #ipfs
<kaoD>
okay, I think this question is better: how does libp2p compare to Telehash?
<Kubuxu>
stoogenmeyer: it doesn't get distributed when you add it, it is available to anyone that has the hash. I don't know its hash until I add it, and if I have it locally it is better for me to add it than to download it from you.
<Kubuxu>
kaoD: Telehash is meshnet/overlaynet protocol. libp2p is library for direct p2p communication having multiple useful primitives built in
<Kubuxu>
Swarm and peer discovery, NAT punching, multiplexing streams and so on.
<stoogenmeyer>
Kubuxu: ok I think I get it.. so locally adding it is effectively the same as getting it from someone else using the correct hash
<stoogenmeyer>
thanks for the explenation
<Kubuxu>
yup
<kaoD>
thanks Kubuxu
<Kubuxu>
no problem
<freemint_>
Is there some software for IPFS, which implements Access controll, VPNs and infrastructures which have a single point of trust?
<Kubuxu>
freemint_: no there is not, yet. There is plan for it.
<freemint_>
Furthermore is there a way for enteties to censor a content?
<Kubuxu>
There is "bad bits" list that hasn't seen much use, it is a way for a gateway providing entity to censor the content.
<Kubuxu>
The content in the network itself cannot be censored but you store only things that you explicitly requested.
ashark has quit [Ping timeout: 272 seconds]
<freemint_>
Ok, so a boarding school/ mallicous regime would provide gateways for the pupil/people, they have to use to get online.
<Kubuxu>
Yup, they could do that.
Akaibu has quit [Quit: Connection closed for inactivity]
<Kubuxu>
but nothing prevents users to try circumvent the limitations and just connect to the network directly.
<Kubuxu>
as they do now.
<freemint_>
Ok, could they enforce that people use their keypairs?
<freemint_>
(in case of non circumvention)
<Kubuxu>
In the HTTPs CA Root style, no.
<freemint_>
So they would most likely only give access to the staic web and a few selected sites they curate.
<freemint_>
*static
<Kubuxu>
I have no idea, what I know is that skilled circumventing would be much easier.
<Kubuxu>
now you always need to connect to the source, with IFPS anyone can be the source.
<Kubuxu>
So you could just move dups on HDDs to circumvent.
<freemint_>
Is it possible for a malicous entity to create a cyclic chain of public keys signing each other such that any try to resolve them results in an DDOS?
<freemint_>
(When somebody tries to resolve they are stuck in a loop.)
<Kubuxu>
There is depth limit to IPNS resolution.
<Kubuxu>
something along 200
<freemint_>
Ok, better be safe around nodes with pointers :D
<freemint_>
Cyclic directories in linux filesystems are a thing\^{TM}, can you preserve that when hosting them over IPFS?
<freemint_>
Since they can not all be immutable ...
<freemint_>
How is hash collision handled, i found nothing in the repo about that.
Oatmeal has quit [Ping timeout: 264 seconds]
cketti has quit [Ping timeout: 246 seconds]
dguttman has joined #ipfs
<r0kk3rz>
freemint_: with the function used a collision is exceedingly unlikely
<freemint_>
I do not think sha256 is a likely canidate, but when people can ship there own has function s**t is going to come down, like a hash function which just takes the first byte.
mpi has joined #ipfs
<freemint_>
In the stanford seminar one semingly important person said you could use any hash function
ylp has quit [Ping timeout: 260 seconds]
<freemint_>
*important seeming
kaoD has quit [Quit: Leaving]
<jonnor>
freemint_, the identifier used for hashes is a 'multihash', meaning it contains a specification of the hash function used https://github.com/jbenet/multihash#format
<jonnor>
this is intended to make it easy to change hash functions. Plug in a new function where handling the multihash, and good-to-go
ylp has joined #ipfs
<jonnor>
I don't know which hash functions an IPFS must support right now to be considered fully implementing the spec though
<freemint_>
ok, is there a query you can make if you encounter an unknown hashfunction id to get an hash function implementation for your platform?
mpi has quit [Read error: Connection reset by peer]
mpi has joined #ipfs
<jonnor>
not that I'm aware of, and that sounds like a huge security risk
matoro has quit [Ping timeout: 250 seconds]
matoro has joined #ipfs
<freemint_>
Thats what you have the web of trust like signing for, rigth?
mpi_ has joined #ipfs
<freemint_>
to mitigate trust issues..
<jonnor>
sure, but always better to not need trust at all
<freemint_>
Well it would be opt-in anyway.
mpi has quit [Ping timeout: 258 seconds]
<freemint_>
Is it possible to delete content? There is a distributed approach to self deleting files. Is anything like that possible in IPFS
dguttman has quit [Quit: dguttman]
<r0kk3rz>
freemint_: ipfs doesnt really do 'delete'
<freemint_>
Thats a drawback.
<r0kk3rz>
it can forget, but not on command
<freemint_>
It would be nice if there was an send "a please forget that content" to any nodes holding that content
dguttman has joined #ipfs
Oatmeal has joined #ipfs
<freemint_>
The node can forget what ever he wants, but one could also punish nodes reputation which serve asked to be deleted data. It would be in the nodes interest to respect the right to be forgotten.
<jonnor>
would only serve censorship of large actors?
<freemint_>
If it is tied to public private key infrastructure it would only serve self censorship.
dguttman has quit [Client Quit]
<Kubuxu>
freemint_: what you are describing is something that IPFS on purpose decides not to touch on, also anyone could say that "yeah I deleted that content"
<freemint_>
He had to prove that he uploaded it first
<Kubuxu>
Anybody that has that file can say that.
<freemint_>
You could tie it to the hash function, detetable date is uploaded with the sha256 of a particular value only the uploader knows,
<freemint_>
*deletable data
<freemint_>
he can use that value to legitimize his request.
<Kubuxu>
It could be implemented as an upper, as you said WoT like layer.
<Kubuxu>
or federated
jedahan has quit [Quit: My Mac has gone to sleep. ZZZzzz…]
<freemint_>
I try to find a solution for the somebody uploads an picture that could be used against him(blackmailing/bullying) on an ipfs backed social network by accident/drunkeness
<freemint_>
- problem
<Kubuxu>
Then someone takes this data, reuploads it and then you the system doesn't work.
<freemint_>
If it has the same hash it does
jedahan has joined #ipfs
<Kubuxu>
It would have other hash.
<Kubuxu>
Like you can reencode the picture.
<freemint_>
I agree with that, but than it is no longer his picture just a picture of him and he has no legal claim to get it removed
<freemint_>
Ofcourse you guys could cry for natural selection and person ends up killing himself, but that would give IPFS a very bad taste.
Penta has joined #ipfs
<Kubuxu>
The same can happen in regular Internet.
<Kubuxu>
There is nothing you can do.
Pent has quit [Ping timeout: 250 seconds]
brabo has quit [Ping timeout: 260 seconds]
<freemint_>
I agree in that case if somebody made a copy an distributes it in print or transformed and reuploaded all bets are off.
brabo has joined #ipfs
brabo has quit [Remote host closed the connection]
<freemint_>
What arguements do you have against an implementation of such a feature?
<Kubuxu>
It is extremely easy to work around, could also lead to censorship and potential abuse.
<freemint_>
IT could NOT be used for censorship(other than self-censorship) UNLESS SHA256 is brocken.
ljhms has joined #ipfs
<Kubuxu>
No it can be used, just "negotiate" with author enough.
<freemint_>
Just make the wallet rubberhose cryptography resistant.
<r0kk3rz>
Kubuxu: ipfs might have to deal with the whole 'right to be forgotten' thing in the EU though
<r0kk3rz>
since its, you'know, the law n stuff
<r0kk3rz>
it does have its practical limitations for sure thoug
<freemint_>
So if you provide an opt-in solution for deletion which is enforced by the whole network, you did what you can do to archive that
brabo has joined #ipfs
<Kubuxu>
'right to be forgotten' and DMCA are addressed to a entities, like websites operators and so on. The website operator will remove the content but there is no way that network will forget, same with the internet.
* A124
thanks deltab, helped. I did found out the Base, in the meantime, not the alphabet.
<freemint_>
That is the state with the current internet. But yeah, since we are complettly rewriting it there is no way we can make it a better place.
<Kubuxu>
"remove" call in the network wouldn't work, first thing, I think, people would do if it gain any controversy (like gov forcing someone to remove the content) would be releasing builds that don't follow it.
<Kubuxu>
s/gain/have gained/
<A124>
It would be disaster.
<freemint_>
I agree, but these builds would suffer with reputation until more then 50% of the nodes are against that feature.
<freemint_>
It would be a democracy.
<Kubuxu>
No it wouldn't be. It is enough to have one node that still keeps the content.
<Kubuxu>
It isn't blockchain.
cemerick has joined #ipfs
PrinceOfPeeves has joined #ipfs
<freemint_>
This node would get a very bad reputation until he is kicked out/served with to low priority to be usefull, but if the mayority of the clients choose the compiler flag option with no rigth to be forgotten, then the networks shifts.
<Kubuxu>
But nodes have no reputation system.
<freemint_>
Oh sorry, is that a planned and not-yet implemented feature, since i am pretty sure i read/heared about it.
<Kubuxu>
Also I would love to see reputation system but not inside of IPFS directly, it can be built on top as a application or something and I am still to see reputation system that can't be sybiled.
ygrek has quit [Ping timeout: 276 seconds]
<freemint_>
I would be afraid if there was one. What is good is determined by the mayority of votes.
jedahan has quit [Quit: My Mac has gone to sleep. ZZZzzz…]
<freemint_>
Any reputation system than can not be sybiled can not be naturally selected out of use.
<freemint_>
*Is not subject to competition.
jedahan has joined #ipfs
<freemint_>
Kubuxu: can you agree with that notion that sybility is a quality necessary for competition and that competition between different system is necessary in the long run?
<r0kk3rz>
plus the whole tyranny of the majority, groupthink issues that arise from having a perfect non-sybill reputation system
<Kubuxu>
As potential user I see sybility as a problem, if WoT can be sybiled then it is useless.
<Kubuxu>
r0kk3rz: nothing says you have to have just one WoT, you could have purpose specific WoT
<freemint_>
Ok, we seem to have disagreeing opinions on that one, but my opinion seems hte unpopular one, so i stay silent.
<freemint_>
about it
<r0kk3rz>
yeah thats true i guess, just like real life
<Kubuxu>
What is gained by sysbility, if network can be sybiled then anyone can be the most powerful entity, what is the purpose of that network then, who can sybil the network better?
<freemint_>
There is proof of work.
<freemint_>
There is proof of stake. Not any entety can be the most powerfull one.
<Kubuxu>
Then it isn't WoT, it is Web of who has most money.
<freemint_>
I draw back my case for sybility, maybe the wars which protocol shall be used must not be fought out at protocoll level.
<freemint_>
But i think the please delete request can be implemented on top of a WOT network.
zorglub27 has joined #ipfs
brabo has quit [Ping timeout: 260 seconds]
<freemint_>
The main motivation for linking it to the reputation is that all the nodes which are indifferent to the content have an interest in deleting it.
Penta has quit [Quit: Leaving]
<freemint_>
If it adds some peer pressure on the advessary it is fine as well.
Pent has joined #ipfs
<r0kk3rz>
freemint_: so the network will have to maintain a hash blacklist?
<freemint_>
I do not know, i fear that would be the case, but there could be a neat trick around it.
<freemint_>
I have not enough experience with distributed systems to deny or agree with that statement.
<freemint_>
r0kk3rz: For the sake of the arguement lets assume yes.
matoro has quit [Ping timeout: 260 seconds]
brabo has joined #ipfs
<r0kk3rz>
you could blockchain it i guess, it wouldnt be too bad but still adding a whole new layer of complexity to manage 'deletion'
<r0kk3rz>
but i would imagine some kind of delete command will be put in place for ipfs-cluster / filecoin
<r0kk3rz>
even if its not authoritive and up to each node to actually delete it
<freemint_>
Is a few life safed worth the computational, adoption and development overhead? Only time will tell
mpi_ has quit [Remote host closed the connection]
<freemint_>
for clusters it would make sense.
<r0kk3rz>
its a cost of the highly public world we live it, it takes very little action to completely ruin someones life
<r0kk3rz>
and i dont think trying to fight the streissand effect is going to change that
<freemint_>
I get your point, it is a decission the community needs to make
<freemint_>
There is no obvious rigth or wrong
zorglub27 has quit [Ping timeout: 276 seconds]
<freemint_>
Once an advessary has the data it is over. This process would change the behavior of the bystanders in a way that they do not provide the advessary with the information necessary to harm the victim.
matoro has joined #ipfs
<freemint_>
Does the community think this is something worth getting an issue on GitHub?
s_kunk has quit [Ping timeout: 240 seconds]
stoogenmeyer has quit [Remote host closed the connection]
ashark has joined #ipfs
<r0kk3rz>
on the flip side, i think the generation growing up today will treat this very differently. since they will all have had their whole life online from the day they were born
<r0kk3rz>
and everyone does stupid shit growing up, or things they might regret later
<freemint_>
I am from that generation and try to rig the system in my favor :D
brabo has quit [Ping timeout: 260 seconds]
matoro has quit [Ping timeout: 260 seconds]
<freemint_>
If you do not have that mechanism, it would mean there would be something like an age restriction on posting stuff.
<r0kk3rz>
i was more meaning the kids my friends are having, and literally posting pictures of daily online
<r0kk3rz>
but sure, social media has been around long enough for whatever effects to occur
<freemint_>
But there they can ~delete~ it.
<freemint_>
bbl
brabo has joined #ipfs
stoogenmeyer has joined #ipfs
<ipfsbot>
[go-ipfs] whyrusleeping pushed 1 new commit to master: https://git.io/vof8w
ZaZ has quit [Read error: Connection reset by peer]
cketti has joined #ipfs
luigiplr has joined #ipfs
zorglub27 has joined #ipfs
palkeo_ has quit [Ping timeout: 264 seconds]
Foxcool has quit [Remote host closed the connection]
<jonnor>
I did 'ipfs add' on a file, and got the IPFS hash "QmdxVDmu3ZzzXTsYMsANpf5px89kiHHEUHtHw4vcaNAK6S". I then tried to use ipfs.chunk.get of this hash. But the response I get has 8 bytes of padding in front, and 3 after? So I tried ipfs.object.get and then I just get "Error: not valid multihash". What am I doing wrong?
Foxcool has joined #ipfs
<xelra>
freemint_: If no node has the data anymore, the data is not available anymore. So it's deleted.
<Kubuxu>
jonnor: are you sure that you copied the hash correctly?
<Kubuxu>
or if you are using js-ipfs check if you don't have to specify encoding somehow
<apiarian>
what is the life cycle of an object that i add to my node? does it hang around forever, or does it get garbage-collected at some point if i don't explicitly pin it or something like that? specifically asking about objects that i create with my node, not objects that i pull in from the swarm
<jonnor>
I'm using js-ipfs-api with go-ipfs running locally
<Kubuxu>
jonnor: check if you copied it correctly
<Kubuxu>
apiarian: ipfs add hash --pin=true by default
<Kubuxu>
so added file is pinned while adding
<A124>
Is publish recursive? Does it cache the contents of the three localy?
<Kubuxu>
you can do `ipfs add --pin=false file` to not pin it
Encrypt has quit [Quit: Quitte]
<Kubuxu>
A124: you mean `ipfs name publish`?
<A124>
Yes.
<Kubuxu>
ipfs name publish only updates the IPNS pointer
<A124>
"If you ran that on the same machine, it should return instantly, as you have cached the entry locally; give it a shot on another computer running ipfs."
<apiarian>
Kubuxu: is that also true for ipfs object new/patch things?
<Kubuxu>
I think no
<freemint_>
back
<apiarian>
ok, so that seems sensible
<jonnor>
Kubuxu, yeah I did ipfs get on the hash I got in the error message. Tested again now, it fetches and is valid file with expected content
<A124>
Kubuxu: So then this a) confuses me, b) trying to find way to "synchronize" local stuff based on ipns, while purging old stuff (that the ipns no longer points to).
<Kubuxu>
jonnor: you must have copied the hash wrong then, if you got `not valid multihash`.
<Kubuxu>
A124: have a file with old hash (lets name it `old`), ipfs name resolve /ipns/$ID > new; ipfs pin add $(cat new); ipfs pin rm $(cat old); mv -f new old
<jonnor>
can the hash contain non-ascii characters or other invisible things?
<Kubuxu>
No, it is base56-bitcoin iirc
<A124>
err.. 58?
<jonnor>
h err QmdxVDmu3ZzzXTsYMsANpf5px89kiHHEUHtHw4vcaNAK6S Error: not valid multihash
<jonnor>
I then check: ipfs get QmdxVDmu3ZzzXTsYMsANpf5px89kiHHEUHtHw4vcaNAK6S
<A124>
Someone earlier today did say it is Base58.
<A124>
Kubuxu Thank you. If it is recursively, that would remove some new stuff also, unless I am wrong.
<Kubuxu>
A124: ipfs pin only saves the root in recursive version
<A124>
So I would have to go along the three and get all hashes, diff, and remove old, or would there be simpler way? ... Yes I know, unless you specify recursively, which is what I want.
<Kubuxu>
No it is by default recursive, it saves the root and if garbage collection is performed, it expands that root to save all sub-hashes from the collection.
redfish has joined #ipfs
redfish has quit [Client Quit]
redfish has joined #ipfs
zorglub27 has quit [Ping timeout: 276 seconds]
cemerick has quit [Ping timeout: 260 seconds]
<A124>
Oh. So I would get the expected behaviour, correct? Thanks again.
<jonnor>
Kubuxu, looked at the code... Default enc is 'json', so I need to specify enc: 'base58'
<sydbarrett74>
A124: i wasn't trying to sound off-topic...i was just looking for someone who is using ipfs and sfw in conjunction :)
corvinux has joined #ipfs
corvinux has quit [Remote host closed the connection]
corvinux has joined #ipfs
<A124>
sydbarrett74 It shares through federation, composes by refactoring and wraps data with visualization. ... Well I got no idea what that even means. Not running it, but interested.
corvinux has quit [Remote host closed the connection]
<sydbarrett74>
i think it would be ideal for running atop ipfs..afaik, swf in its default implementation suffers from fragility because it's still built atop http
chriscool1 has quit [Ping timeout: 260 seconds]
s_kunk has quit [Ping timeout: 244 seconds]
<A124>
sydbarrett74 Of course, Looking at how it works. Ain't bad. Know any examples of production sites that run on top of that?
<sydbarrett74>
alas, no :(
<sydbarrett74>
i'm sure there must be
s_kunk has joined #ipfs
<A124>
I am getting currently more familiar with ipfs innards and interfaces, extending the javascript to be able to use ipfs via API whould not be that bad. Also the question is if one would want multilinks, and fallback to some gateways. Definitely him me again if you keep interested, sydbarrett74
<sydbarrett74>
oh i surely will, because once i get a beefy desktop machine built instead of playing with stuff on my poor underspec'd laptop, i'd like to start playing with all this more seriously
<A124>
sydbarrett74 I can help you even pick components for your machine if interested I was building stuff, including for film studios. It is about balance of what you get. If you are on Linux SSD+HDD + b-cache gives performance and lot of storage.
<sydbarrett74>
the biggest part is affording it :/ but i'd like to have a list to aspire to..i want something i can use as a server but also a decent gaming rig
<sydbarrett74>
and i'd like something that i can throw up to 64GB of RAM into
corvinux has joined #ipfs
<sydbarrett74>
afk
cemerick has quit [Ping timeout: 250 seconds]
jonnor has joined #ipfs
jedahan has quit [Quit: My Mac has gone to sleep. ZZZzzz…]
* A124
to apiarian: you know that you can actually create your own service? Above corenet, so you do not have to rely on files.
stoogenmeyer has quit [Ping timeout: 244 seconds]
<apiarian>
A124: i looked at that, but back when i was designing things it ended up recreating a lot of the IPFS plumbing, so i decided to build it as an ipfs-api client. i want to reuse my actual ipfs node.
rendar has quit [Quit: std::lower_bound + std::less_equal *works* with a vector without duplicates!]