<Mateon1>
Ugh, apparently Python is not fit for processing massive text files
<Mateon1>
I really wanted to avoid writing C, but I can't now
<Mateon1>
Wait, nevermind... I am using linear search for the whole file, because I'm stupid and "end - start < MIN_LINEAR" differs from "end - start > MIN_LINEAR"
<Mateon1>
Oh my god, it is near instantenous now
<dignifiedquire>
:D
<dignifiedquire>
not only the sword has to be sharp, you also need to know how to wield it :)
<Mateon1>
Still, C would be a couple times faster
<Mateon1>
After that, I'd be IO bound
galois_d_ has quit [Remote host closed the connection]
galois_dmz has joined #ipfs
pfrazee has joined #ipfs
wrouesnel has quit [Remote host closed the connection]
<Mateon1>
WHOO, it works!
<Mateon1>
I am getting all the root hashes in my repo
<Mateon1>
Root, as in, not pointed to from any other hash
<Mateon1>
I'm seeing lots of images from ipfs.pics
martinBrown has quit [Ping timeout: 256 seconds]
pfrazee has quit [Ping timeout: 240 seconds]
<Mateon1>
More bugs! on `ipfs ls <hash>`: Error: proto: bad wiretype for field unixfs_pb.Data.Type: got wiretype 2, want 0
wrouesnel has joined #ipfs
<Mateon1>
I'm going to sleep for now, gonna test a bunch more tomorrow
stevenaleach_ has joined #ipfs
stevenaleach has quit [Read error: Connection reset by peer]
wezaui has quit [Read error: Connection reset by peer]
Boomerang has quit [Quit: Lost terminal]
jkilpatr has quit [Ping timeout: 255 seconds]
maxlath has quit [Quit: maxlath]
faenil is now known as faenil_
jkilpatr has joined #ipfs
jkilpatr has quit [Ping timeout: 240 seconds]
jokoon has joined #ipfs
maxlath has joined #ipfs
victorbjelkholm has quit [Quit: ZNC - http://znc.in]
victorbjelkholm has joined #ipfs
jkilpatr has joined #ipfs
taaem has quit [Remote host closed the connection]
mguentner2 is now known as mguentner
taaem has joined #ipfs
__uguu__ has quit [Read error: Connection reset by peer]
__uguu__ has joined #ipfs
taaem has quit [Remote host closed the connection]
uktjames has joined #ipfs
<uktjames>
I want to publish a website using IPNS. However, If I update the website later on, how to ensure the link the user has won't break?
faenil_ is now known as faenil
cemerick has quit [Ping timeout: 248 seconds]
cemerick has joined #ipfs
<victorbjelkholm>
uktjames: IPNS always stay the same (considering you have the same IPFS configuration with the keypairs)
<victorbjelkholm>
so you'll update what your ID points to (in IPNS), a different IPFS hash for example, and people can still resolve with the same IPNS name
CmdSft[m] has joined #ipfs
CmdSft[m] has left #ipfs [#ipfs]
anonymuse has joined #ipfs
chovy has quit [Ping timeout: 240 seconds]
espadrine_ has quit [Ping timeout: 264 seconds]
<uktjames>
victorbjelkholm: Yes but isnt ipns - peer-id/IPFS hash? So If I change the website, and the hash changes then how will the user exactly view it?
tabrath has joined #ipfs
<victorbjelkholm>
uktjames: no, you'll point IPNS to the hash, without including the actual hash when linking to IPNS
<Mateon1>
I'd wish to look if there are new things on there, but my Chrome is misbehaving, and doesn't want to fork
<Mateon1>
All current tabs work perfectly, and I can duplicate some tabs, but I can't create new tabs
tmg has quit [Ping timeout: 240 seconds]
john4 has joined #ipfs
john3 has quit [Ping timeout: 276 seconds]
anonymuse has quit [Remote host closed the connection]
Tsutsukakushi has quit [*.net *.split]
Tsutsukakushi has joined #ipfs
anonymuse has joined #ipfs
arpu has quit [Ping timeout: 276 seconds]
mildred3 has quit [Ping timeout: 240 seconds]
mildred3 has joined #ipfs
arpu has joined #ipfs
ulrichard has quit [Remote host closed the connection]
ylp has quit [Quit: Leaving.]
cemerick has quit [Ping timeout: 248 seconds]
cemerick has joined #ipfs
<kythyria[m]>
Hm, will anything break if my .ipfs directory is accessed by instances on different machines?
cemerick has quit [Ping timeout: 248 seconds]
ylp has joined #ipfs
G-Ray_ has quit [Quit: G-Ray_]
G-Ray_ has joined #ipfs
<Kubuxu>
kythyria[m]: there is a lock file in there
<kythyria[m]>
To prevent concurrent access?
ygrek has joined #ipfs
<kythyria[m]>
So I guess that's a "it would if the lockfile wasn't there"
<achin>
my guess is that the network is not designed to handle two nodes running at the same time, each claiming to have the same ID
G-Ray_ has quit [Quit: G-Ray_]
Encrypt has quit [Quit: Quit]
<Kubuxu>
FrankPetrilli: so about kubernetes-ipfs, I am thinking about introducing RPC server, that would be main process started by the kubernetes, then it would start stop and manage ipfs instance within. What do you think about it?
G-Ray_ has joined #ipfs
<A124>
richardlitt: Can someone solve https?
<A124>
For the discourse.
<richardlitt>
A124: What do you mean by solve?
<richardlitt>
Set it up? yes. What We need to do there is to switch to an actual instance, and not a trydiscourse.com instance
<richardlitt>
lgierth: How easy would that be? I think you looked into this.
ygrek has quit [Ping timeout: 245 seconds]
john1 has joined #ipfs
john4 has quit [Ping timeout: 260 seconds]
galois_d_ has joined #ipfs
john2 has joined #ipfs
john1 has quit [Ping timeout: 256 seconds]
galois_dmz has quit [Ping timeout: 245 seconds]
palkeo has joined #ipfs
faenil is now known as faenil_
neurrowcat has joined #ipfs
s_kunk has quit [Ping timeout: 260 seconds]
robattila256 has quit [Ping timeout: 276 seconds]
cemerick has joined #ipfs
gully-foyle has joined #ipfs
faenil_ is now known as faenil
s_kunk has joined #ipfs
ygrek has joined #ipfs
arkimedes has joined #ipfs
G-Ray_ has quit [Quit: G-Ray_]
vapid has joined #ipfs
<vapid>
what does
<vapid>
being
<vapid>
Unavailable for Legal Reasons
<vapid>
entail?
<vapid>
spooky
<whyrusleeping>
vapid: means that we had a DMCA that we have to respect on the ipfs.io gateways
<vapid>
ah right
<whyrusleeping>
but yeah, we make it spooky on purpose
<whyrusleeping>
that sort of thing shouldnt just 404
tabrath has quit [Ping timeout: 258 seconds]
espadrine has quit [Quit: Konversation terminated!]
<Stskeeps>
whyrusleeping: does ipfs have the dmca blacklist functionality yet ooi?
<Stskeeps>
was still getting planned last i saw, admittedly long ago
<whyrusleeping>
yeah... its still in the "not a high enough priority" phase right now
espadrine has joined #ipfs
bastianilso_ has joined #ipfs
draynium has quit [Quit: <No reason given>]
tabrath has joined #ipfs
draynium has joined #ipfs
anonymuse has quit []
espadrine has quit [Quit: Konversation terminated!]
seharder has joined #ipfs
gavinmcdermott has joined #ipfs
espadrine has joined #ipfs
structur_ has joined #ipfs
ralphthe1inja has joined #ipfs
structuralist has quit [Ping timeout: 255 seconds]
ralphtheninja has quit [Ping timeout: 240 seconds]
<gozala>
I’m afraid I don’t have enough context to provide useful feedback
<gozala>
daviddias: I’m happy to outline the problem I’ve discussed with jbenet
<gozala>
daviddias: and in fact since that conversation I investigated further into issue and I can tell with confidence that it won’t be easy to make firefox use hash as origin
<gozala>
assuming he fs://ipfs/${hash}/path/to style URLs will be used
<kythyria[m]>
Well, if you're dead set on violating the structure of URIs, yes.
dbri1 has joined #ipfs
dbri has quit [Write error: Broken pipe]
<gozala>
daviddias: jbenet I think your best bet would be to either have origin that matches the whole url (without query and #)
<gozala>
that I have hard time believing would be visible sell to Mozilla or browser vendors
<gozala>
or alternatively share same ‘ipfs’ origin across all content
<gozala>
in which case it won’t actually work correctly
<gozala>
unless you guys have well thought, different content policy solution
<gozala>
but even then I imagine some resistance due to user expectations etc...
<gozala>
kythyria[m]: can you please elaborate ?
<kythyria[m]>
The origin is supposed to be the part between the // and the next slash.
<gozala>
kythyria[m]: I feel like you’re not totally following what I’m trying to say
<kythyria[m]>
(well, technically it's the part before the third slash since the scheme matters)
espadrine_ has joined #ipfs
<gozala>
kythyria[m]: what I’m saying is the way firefox (at least) implements content security policy is heavily based on origins that as you say are pretty much hostname
<kythyria[m]>
And in the everything-is-a-unix-path thing IPFS uses, the `/ipfs/` is more like a scheme.
<gozala>
kythyria[m]: there are support of simple-uri stuff in firefox in which case content security policy code uses whole url (without query and #) to assert privileges
<gozala>
instead of origin
<gozala>
neither option is going to work for IPFS
<gozala>
so only option would be to add new type of nsIURI implementation (that can only be done in C++ and has to be shipped with Firefox)
<kythyria[m]>
Why wouldn't it work?
<gozala>
and also patch nsIScriptSecurityManager
<gozala>
kythyria[m]: In other words you have two options:
<gozala>
1. Everything coming from IPFS shares same origin - ‘ipfs'
gigq_ has quit [Quit: leaving]
<gozala>
2. Everything coming from IPFS get own unique origin
<gozala>
With No 1, it goes against current content security policy of the browsers & would be very hard to sell to any vendor, without compelling alternative story.
<kythyria[m]>
Or, you could create a pair of new schemes, `ipfs://` and `ipns://`
<gozala>
And even with compelling story I have my doubts
<gozala>
kythyria[m]: yes so that has being discussed in the past and decided agains in favor of `fs://ipfs/`
<kythyria[m]>
Why?
<gozala>
kythyria[m]: but ignoring that there is an issue of hostname case insensitivity
<gozala>
while hashes are case sensitive by default
<gozala>
kythyria[m]: so pretty much only viable option is to:
<kythyria[m]>
It's probably an easier sell to get them to add the option for case sensitivity.
<gozala>
kythyria[m]: There’s a w3 spec on how to parse url’s
structur_ has quit [Remote host closed the connection]
<gozala>
effort to alter it is greater than you may think
<gozala>
kythyria[m]: Anyway so only option I see is
<gozala>
kythyria[m]: Unless I’m mistaken that is doable in firefox without changes to the gecko or specs
<gozala>
unfortunately not in Electron though
* kythyria[m]
wonders if it'll be easier to get browsers to add case insensitivity and add it to the spec after the fact, than to get them to allow URI schemes to define their own notion of "origin".
<gozala>
kythyria[m]: but even then it’s not going to be ideal because if someone just takes {hash} and adds `ipfs://{hash}` to it’s not going to work
<kythyria[m]>
Redirect if the origin is `ipfs`?
<kythyria[m]>
(granted, that'll probably still screw up the notion of origin)
<gozala>
if you do that it will work with existing content policy as desired
structuralist has quit [Ping timeout: 255 seconds]
<muvlon>
case insensitivity is a complete mess for non-ascii urls
mildred3 has quit [Ping timeout: 248 seconds]
<gozala>
kythyria[m]: let me put it this way, browser implementers are very conservative about changes to code that affect any of the security code. Due to possible effects it may have on millions of users privacy
<gozala>
changes to the way browser parse urls or handle content security is going to be a very hard sell
<gozala>
and to be honest it’s not like browser vendors are sold on shipping ipfs in browser either
<kythyria[m]>
So basically you're only stuck with awful choices.
<gozala>
so the less changes you require the better chances are to get ipfs in browser
<gozala>
kythyria[m]: so for one ipfs needs to have a story how content security is supposed work
<gozala>
and how that interacts with existing system or content origin policy
rendar has quit [Quit: std::lower_bound + std::less_equal *works* with a vector without duplicates!]
<gozala>
kythyria[m]: if you want to convince browser vendors to change the way security works or urls work
<gozala>
well you better have a elaborate explanation why ipfs can’t change instead to be compatible with existing system
<gozala>
for one it’s a lot less risky to make changes on ipfs side than on browser side
<A124>
I got a problem, ipfs says needs migration, while migration tools says alerady at version
<A124>
Using master versions of both ipfs an migration tool.
<kythyria[m]>
gozala: Huh, yeah, it does case insensitive comparison.
Encrypt has quit [Quit: Quit]
<gozala>
kythyria[m]: I understand a sentiment a noble cause in trying to fix old mistakes, but unless there is some gradual plan, meaning intermediate steps to get there, I have my doubts it’s going to succeed
<kythyria[m]>
"Have the option for URIs to be case sensitive on a per-scheme basis" isn't a huge step?
<A124>
Using 0.4.4 release tarball same thing.
<gozala>
kythyria[m]: depends what you thing huge step is
<gozala>
kythyria[m]: If I had to guess, to get IPFS into firefox it would have to be shipped separately as system add-on first
<gozala>
to study what it buys us
<gozala>
pros / const
<gozala>
cons*
<gozala>
and assuming it was a visible success for Mozilla mission
<gozala>
we would work towards integrating it into source code
<kythyria[m]>
There's probably also no way to make a scheme that takes over from a specific URI prefix.
<gozala>
kythyria[m]: I’m not following that last one
muvlon has quit [Quit: Leaving]
<kythyria[m]>
For compatibility with people that act as if they think the scheme prefix for IFPS is `https://gateway.ipfs.io`
<gozala>
kythyria[m]: again that seems like change completely how browser works
<kythyria[m]>
Though... IPFS' insistence on _everything_ being a unix path means you probably end up needing custom definitions of "origin" anyway.
<gozala>
kythyria[m]: consider Mozilla’s position
<gozala>
why ?
<kythyria[m]>
So that people can continue to write the One True Path Format and have it work.
<gozala>
kythyria[m]: no why ship ipfs and do this hacks in first place ?
tmg has joined #ipfs
<gozala>
why not competing solution that would work out of the box for instance ?
<gozala>
is ipfs somehow serves Mozilla mission ?
<A124>
_competing_
<A124>
Show me one, or all.
<A124>
It's close to saying why use HTTP + Browser + JS + CSS + HTML instead of competing solution that does it all?
<A124>
At least in my eyes.
<gozala>
A124: I’m not really trying to argue here, what I’m trying to say is
Mizzu has quit [Ping timeout: 255 seconds]
<gozala>
the fact is it’s not clear that ipfs is something that serves Mozilla’s mission
<gozala>
and even then it’s not clear it’s worth putting effort into it vs some other things
<gozala>
unless it’s absolutely clear not just for few ppl in this channel
<gozala>
it’s very unlikely to get major changes landed
<gozala>
at best I imagine if contributors would submit a patch to implement new nsIURI implementing component
<A124>
Well, it is not even halfway, so that is reasonable.
<gozala>
that is treated differently and is prefd out by default
<gozala>
and then it is used by ipfs to do it’s thing it maybe visible path to pursue
<A124>
Well imagine you were building custom made car. And you argue why to drive that car even.
<gozala>
and even then you’d be asked why those changes ?
<gozala>
A124: you don’t have to convince me that IPFS is good fit for Mozilla
<gozala>
I’m here because I believe so
<gozala>
but there needs to be success story to build upon and to support that claim
<A124>
Agreed. Well in the first place I did not understand what the fuss was about.
<gozala>
and it’s hard to get there if you try to make invasive changes solely on the promise of success story
<A124>
Elon Musk.
<gozala>
Anyway I chatted to much, I need to get back to my actual work
<kythyria[m]>
Wouldn't it have to be `fs:/ipfs/` though?
<Kubuxu>
gozala: if you do it, do two things, 1. use one slash instead of two, 2. prefix with /ipfs/ either way
<kythyria[m]>
Otherwise things that assume you start an absolute URI `/ipfs/` will be interpreted by a browser as `fs://ipfs/ipfs/...`
<gozala>
Kubuxu: from what I remember idea was to support 1-3 slashes
<Kubuxu>
if you do `ipfs:/QmAAA` it won't be compatible with most of the sites
<Kubuxu>
because people might type them in
<Kubuxu>
but it should redirect them to single slash
<Kubuxu>
same way you can type in 4 slashes after http: in your browser
<gozala>
Kubuxu: no ipfs: protocol would have to be internal implementation detail
<gozala>
ipfs://{hash_in_base16}/path/wih/in
<gozala>
and be like above ^
<gozala>
to do content policy properly
<gozala>
we could do whatever for fs: protocol
structuralist has joined #ipfs
<Kubuxu>
I've never seen such a plan
<kythyria[m]>
If you redirect consistently people will see that URL though
<kythyria[m]>
Unless URI handlers can redirect without it showing but still change the origin.
<gozala>
Kubuxu: well that’s my plan not something official
<gozala>
kythyria[m]: yes users will see ipfs://{hash_in_base16}/ indeed
<gozala>
and it won’t be great
xelra has quit [Remote host closed the connection]
<gozala>
Anyway once I’ll get it to work I’ll put it out there so you can tweak it as you like
<gozala>
but at the end of the day users would see ipfs://{hash_in_base16}/path/with-in
<gozala>
there maybe some opportunities to normalize ipfs:{hash} ipfs:/{hash} and ipfs://{hash} and ipfs:///{hash} all into ipfs://{hash_in_base16}
Mizzu has joined #ipfs
xelra has joined #ipfs
<kythyria[m]>
I wonder what will happen with absolute links that are using the unixoid format, though. The browser will want to load ipfs://{hash_in_base16}/ipfs/{hash}
xelra has quit [Remote host closed the connection]
dryajov has quit [Ping timeout: 260 seconds]
xelra has joined #ipfs
xelra has quit [Remote host closed the connection]
cemerick has quit [Ping timeout: 248 seconds]
xelra has joined #ipfs
<lgierth>
not sure what you're trying to achieve -- but the fs: and ipfs: schemes don't have an authority/host part so there's no double-slash
<lgierth>
the path starts right after the colon
<lgierth>
and uri handlers don't have to redirect anymore, they can use some kind of stream content api, at least in firefox
xelra has quit [Remote host closed the connection]
<lgierth>
richardlitt: i haven't looked into https and discourse, but i figure it's possible. i think the last status was that we wanted to switch to a paid account?
<kythyria[m]>
lgierth: The problem is that it needs to be `ipfs://` to make origins work right without changing what's considered an origin.
xelra has joined #ipfs
<lgierth>
ah i see. and then you treat the hash as the authority/host eh
<kythyria[m]>
Yes
<lgierth>
yeah that goes pretty much against all that we have in mind with paths and URIs
<lgierth>
we'll need to figure out a way to still make it work. amendments to the CSP specs, etc.
<lgierth>
we already gave feedback to the suborigins working group, for example
<lgierth>
e.g. every ipfs hash is its own suborigin
<kythyria[m]>
Ah
<lgierth>
your ipfs://hash proposal goes in the right direction in that it acknowledges that every hash is its own origin
<lgierth>
but it throws all the paths work over board :)
<kythyria[m]>
It just doesn't play nice with the IPFS thing where absolute paths start with a single slash.
Zaibon has quit [Ping timeout: 258 seconds]
xelra has quit [Ping timeout: 240 seconds]
Nycatelos has quit [Ping timeout: 252 seconds]
Zaibon has joined #ipfs
Nycatelos has joined #ipfs
eocarragain has joined #ipfs
xelra has joined #ipfs
Captain_Beezay has joined #ipfs
xelra has quit [Remote host closed the connection]
shizy has quit [Ping timeout: 240 seconds]
xelra has joined #ipfs
LeerSpace has quit [Remote host closed the connection]
LeerSpace has joined #ipfs
LeerSpace has quit [Remote host closed the connection]
LeerSpace has joined #ipfs
xelra has quit [Remote host closed the connection]
xelra has joined #ipfs
LeerSpace has quit [Remote host closed the connection]
LeerSpace has joined #ipfs
ashark has quit [Ping timeout: 240 seconds]
ashark has joined #ipfs
LeerSpace has quit [Remote host closed the connection]
LeerSpace has joined #ipfs
lacour has joined #ipfs
ashark has quit [Ping timeout: 248 seconds]
xelra has quit [Remote host closed the connection]
A124 has quit [Quit: '']
xelra has joined #ipfs
robattila256 has joined #ipfs
xelra has quit [Remote host closed the connection]
Caterpillar has quit [Quit: You were not made to live as brutes, but to follow virtue and knowledge.]
xelra has joined #ipfs
<daviddias>
gozala: I hear you and totally understand the requirements and the contraints. Is it clear that cidv1 enables you to have any base you want?
<daviddias>
sprint-helper1 has quit [Remote host closed the connection]
sprint-helper has joined #ipfs
<daviddias>
Also happy to jump on a call and go throughly design details and migration path
<gozala>
daviddias: To be honest I have vague understanding of cidv0 / cidv1 but as far as I understand v1 allows transcoding to arbitrary encodings
Mizzu has quit [Ping timeout: 255 seconds]
<gozala>
daviddias: all I was trying to say I can’t really offer any useful feedback on the thread you asked me to
<gozala>
daviddias: I said that to jbenet as well, I think it is important that you guys (core team) articulate content policy view for ipfs in written form somewhere
<gozala>
that would really help driving conversation
<daviddias>
understood, will draft some of the things i have in mjnd and add to the thread
<gozala>
daviddias: I think there are multiple possible routes to move forward, but unless there’s clarity on what is overall picture it will be hard to converge on anything
<gozala>
daviddias: which thread ?
<gozala>
daviddias: please cc-me or send me a pointer to it
<lgierth>
it'll basically be every-as-an-origin-of-its-own right?
xelra has quit [Ping timeout: 240 seconds]
<lgierth>
*every-hash
<gozala>
lgierth: that’s also my understanding
Nycatelos has quit [Ping timeout: 240 seconds]
<lgierth>
would be useful to have a general list of things-we-need-from-browser-vendors