<codehero>
davidar: you can also look at my config
<tilgovi>
I don't know how well running arch-derivitives works though
<codehero>
i think it's better to just install arch
<codehero>
it's not that hard
<tilgovi>
I don't know enough about arch packages to know what the state of that world is with respect to overlays and such
<tilgovi>
is it easy to manage package sources, and does the community make 3rd-party and experimental packages available anywhere easy?
<davidar>
tilgovi (IRC): I guess mint would also be an option of you didn't want too big of a change
<davidar>
*if
<codehero>
tilgovi: well. there's the aur
<davidar>
tilgovi (IRC): I think AUR is probably the biggest strong point of arch
<tilgovi>
oh right
<tilgovi>
i remember that now
<tilgovi>
(i have looked into these things in the past)
<tilgovi>
you're right
<tilgovi>
that is really awesome
<tilgovi>
god, how in the hell did I ever install kubuntu on a zfs root?
<tilgovi>
I must have done the installation by hand
<tilgovi>
I bet it wasn't that bad. Boot livecd. Drop to terminal. Install zfs packages. Format. Then run the installer and ask it not to format. I could probably manage the same with arch.
<codehero>
yeah. pretty easy, if not easier
<codehero>
you don't have to switch between cli and gui
<tilgovi>
maybe i'll give it a shot soon
<tilgovi>
I really miss the rolling releases from my gentoo days.
<tilgovi>
Every one of the last 3 or 4 upgrades I've done has resulted in failure in the middle of upgrading and some amount of scary rescue effort
<davidar>
@freenode_codehero:matrix.org: how easy is it to install third party stuff on nixos?
<codehero>
nixos is also used to install packages that would usually be installed via cabal, pip, go, ...
<codehero>
it's pretty sweet
<codehero>
i think also npm
<codehero>
and the ruby thingy
<codehero>
oh, also emacs packages
<davidar>
I read that, didn't realise it was that comprehensive though
<codehero>
yupp
<codehero>
but you should wait until security updates are fixed before you use it as your main system
voxelot has quit [Ping timeout: 256 seconds]
pfraze has joined #ipfs
voxelot has joined #ipfs
<davidar>
Ah, what's the story with that?
<whyrusleeping>
someone say something to me?
<davidar>
whyrusleeping (IRC): we were talking about how much arch sucks
<codehero>
davidar: mass rebuilds sometimes cause every package update to be delayed by 1 month or more
<codehero>
including security updates
<codehero>
so, yeah
<codehero>
but pierron is working on fixing htat
<codehero>
*that
<davidar>
@freenode_codehero:matrix.org: ah, how long til that gets fixed?
<whyrusleeping>
davidar: take it back!
<davidar>
whyrusleeping (IRC): lol
<whyrusleeping>
arch is the only distro i've ever used that manages to install graphics drivers without fucking everything up
<davidar>
@freenode_codehero:matrix.org: speaking of, what's the state of graphics drivers on nixos?
<codehero>
they work
<codehero>
i mean. you get the same ones you get on other distributions
<codehero>
installing them is also pretty easy
HoboPrimate has quit [Remote host closed the connection]
domanic has joined #ipfs
<davidar>
cuda?
r04r is now known as zz_r04r
<codehero>
i only have a laptop with intel integrated graphics, but i think cuda also works
<codehero>
now the documentation on nixos isn't always the best....
<codehero>
but the people on #nixos and the nixos mailing list are very helpful
samiswellcool has quit [Quit: Connection closed for inactivity]
<whyrusleeping>
davidar: cuda was the hard part that arch got right for me
<codehero>
heh
simonv3 has quit [Quit: Connection closed for inactivity]
<codehero>
well, the only thing you can do to find out if it really works is trying
<davidar>
@freenode_codehero:matrix.org: ooh, AUR on nixos? :p
<davidar>
whyrusleeping (IRC): cuda is pretty easy on Ubuntu too...
<codehero>
davidar: umm. no. no aur
<codehero>
the package managing works pretty much like AUR though
<codehero>
users just create/update a package and make a pull request
<whyrusleeping>
davidar: it was a bitch and a half to get up to date cuda on ubuntu
<whyrusleeping>
plus, screw actually using ubuntu, unity is an abomination
<davidar>
whyrusleeping (IRC): well, kubuntu
<davidar>
Cuda is actually easier to install than the standalone drivers
<whyrusleeping>
yeah, i just like being able to pick exactly what package i install
<whyrusleeping>
'sudo pacman -S cuda'
<whyrusleeping>
kablam
<davidar>
Yeah, there's a Debian repo for cuda too :p
<davidar>
Anyway, I'll probably try nixos at some point, thanks for answering my questions codehero :)
<ion>
Unity is pretty great.
<davidar>
I don't mind unity aesthetically, but it's stability is a bit crap
<ion>
I like how well you can use it with just the keyboard. It even has a single-keypress cheatsheet for keyboard shortcuts.
<davidar>
I ended up abandoning it because the password field in the lock screen kept disappearing :/
<ion>
And i like the launcher and the window switcher being the same in any system.
<codehero>
davidar: no problem
<davidar>
ion (IRC): it's not too hard to set up kde to act like unity for the most part
<davidar>
But, actually with good performance :p
tilgovi has quit [Remote host closed the connection]
dignifiedquire has quit [Quit: dignifiedquire]
<davidar>
I would have been happy if they just made unity a custom config for kde actually
<davidar>
They're already moving to qt anyway...
<whyrusleeping>
>.> the webui crashed on me
<zignig>
/me passes whyrusleeping the controls for the orbital disintergrator.
<whyrusleeping>
dude
<whyrusleeping>
i dont know whats going on right now
<whyrusleeping>
i also am not webscale, so i dont know how to debug these things
<codehero>
whyrusleeping: :D
<codehero>
same thing happened to me before
<whyrusleeping>
hrm...
<whyrusleeping>
this is troublesome
<codehero>
but i have no idea how to reproduce it
<codehero>
seems to be random
<whyrusleeping>
i want to ship 0.3.9 right meow
<codehero>
:D
<achin>
mrow
<whyrusleeping>
hrm. thats very troublesome
<whyrusleeping>
a bug i thought was fixed is not fixed
Guest73396 has quit [Ping timeout: 260 seconds]
simonv3 has joined #ipfs
patcon has quit [Ping timeout: 240 seconds]
* whyrusleeping
grabs a beer
<whyrusleeping>
bugfixin time
<zignig>
unce unce unce
<achin>
whyrusleeping is invoking the balmer curve
<achin>
ballmer peak
<codehero>
:D
<codehero>
hm. i can't get it to crash, but it doesn't show any node information anymore
<achin>
(protip: doens't work with spelling or typing)
<codehero>
:D
<codehero>
oh well
<codehero>
you can't '>' the core dump.....
<codehero>
or... i'm just an idiot
<codehero>
who knows
<zignig>
codehero: why not ?
<codehero>
should've used 2& i think...
<codehero>
well
<zignig>
yes.... ;)
<lgierth>
hey everyone, just a heads up: i'm about to run `ipfs repo gc` on all ipfs.io hosts, one after another
ygrek_ has quit [Ping timeout: 260 seconds]
<achin>
i'd be curious to know how much data you end up purging (across all nodes)
<whyrusleeping>
whats happening is that the 'log' tab is opening up the ipfs log tail endpoint
<whyrusleeping>
and then not closing it
<whyrusleeping>
so all logs hang trying to write through
<whyrusleeping>
and the daemon halts
<whyrusleeping>
because every goroutine gets stuck in a log
* whyrusleeping
grabs beer #2
<ion>
lgierth: While automatic LRU GC isn’t implemented, perhaps you could schedule a round-robin set of invocations on a different node every day. If some content is popular on the gateways, 7/8 hosts will retain it over any single invocation and it’ll probably get it back from the others before the others are GC’d.
<spikebike>
but all my priceless data!!!!
<spikebike>
;-)
<spikebike>
GC'ing 1/N hosts a day seems pretty reasonable
<spikebike>
ah, yeah, been planning something like that, basically was planning to take N files, query who has them, then add/subtract replication to meet the replication goals
amstocker has joined #ipfs
<whyrusleeping>
anyone know how to make netcat *not* read anything?
<whyrusleeping>
i just want to connect, write something, and then hang without reading
<whyrusleeping>
nikogonzo: i think thats the other way around
<whyrusleeping>
-z says not to write anything
<whyrusleeping>
i want to not read anything
blueskymonster has quit [Ping timeout: 246 seconds]
<whyrusleeping>
spikebike: nc reads and spits to stdout
<whyrusleeping>
that wont work
<lgierth>
whyrusleeping: cat - | nc ?
<lgierth>
should hang until you send ctrl-d
gendale_ has joined #ipfs
<nikogonzo>
whyrusleeping: so you want to start a nc listener, then send data to the listener, but have the listener silently drop whatever came over the pipe? that sounds like nc -d -k to me. with -d you don't even have to redirect stderr/stdout it appears
Eudaimonstro has quit [Remote host closed the connection]
chriscool has quit [Ping timeout: 268 seconds]
Eudaimonstro has joined #ipfs
chriscool has joined #ipfs
<whyrusleeping>
no, i want to send data to an ipfs daemon, and not read anything back
vx has quit [Quit: Connection closed for inactivity]
Guest73396 has quit [Ping timeout: 240 seconds]
<davidar>
whyrusleeping (IRC): kind of like spaghetti sort, minus the spaghetti
amstocker has quit [Ping timeout: 255 seconds]
<davidar>
Or, spaghetti that's been rotated into the fourth dimension :p
<whyrusleeping>
lol
<davidar>
Hehe, I love this sentence: "This is possible because the hand, the spaghetti rods and the table work as a fully parallel computing device. "
<ipfsbot>
go-ipfs/fix/log-hang f21fe5c Jeromy: fix log hanging issue, and implement close-notify for commands...
ipfs_intern has joined #ipfs
s_kunk has quit [Ping timeout: 268 seconds]
harlan_ has joined #ipfs
ilyaigpetrov has joined #ipfs
NeoTeo has joined #ipfs
Zuardi has quit [Remote host closed the connection]
Zuardi has joined #ipfs
dignifiedquire has joined #ipfs
s_kunk has joined #ipfs
dignifiedquire_ has joined #ipfs
dignifiedquire has quit [Ping timeout: 256 seconds]
dignifiedquire__ has joined #ipfs
dignifiedquire has joined #ipfs
dignifiedquire_ has quit [Ping timeout: 240 seconds]
<victorbjelkholm>
spikebike, ipfsbin is not working for you? Would be useful if you sent me the console output if something weird happens
<ianopolous>
davidar: nice bogobogosort :-)
<dignifiedquire>
good morning everyone
<victorbjelkholm>
Good morning!
voxelot has joined #ipfs
voxelot has quit [Ping timeout: 252 seconds]
master8787 has joined #ipfs
rendar has joined #ipfs
stoopkid has joined #ipfs
dignifiedquire__ has quit [Quit: dignifiedquire__]
<NeoTeo>
morning :)
<spikebike>
victorbjelkholm: I added an ipfsbin, then tried to get to it with ipfs cat, took several minutes to get it. A friend couldn't get the website to work which was strange.
ipfs_intern has quit [Quit: Page closed]
<spikebike>
victorbjelkholm: you getting any file handle issues on your node handling ipfsbin?
Guest73396 has joined #ipfs
<victorbjelkholm>
Hm, working fine here. Let me check
<victorbjelkholm>
fifteen minutes ago I got "time="05:45:28:000" level=error msg="err: multipart: NextPart: http: invalid Read on closed Body" module="commands/http""
<victorbjelkholm>
hm, tricky. Oh well, right now is just one node that is pinning the content, in the future, there will be more nodes pinning content, then it's gonna get faster
<victorbjelkholm>
0.3.9
<spikebike>
gotta run, seems to be workign now
<cryptix>
hellow
edcryptickiller has joined #ipfs
edcryptickiller has left #ipfs [#ipfs]
Guest73396 has quit [Ping timeout: 256 seconds]
captain_morgan has quit [Ping timeout: 240 seconds]
harlan_ has quit [Quit: Connection closed for inactivity]
OutBackDingo has quit [Quit: Leaving]
OutBackDingo has joined #ipfs
nessence has quit [Remote host closed the connection]
nessence has joined #ipfs
<victorbjelkholm>
dignifiedquire, how are you finding greekeeper so far? Is a bit noisy, no?
ilyaigpetrov has quit [Quit: Connection closed for inactivity]
<dignifiedquire>
victorbjelkholm: I love it
<dignifiedquire>
it makes it so much easier to keep track of dependency updates and if they are breaking something
<dignifiedquire>
and as it’s a PR I can just review it when I have time
<victorbjelkholm>
yeah, that's true. I just wished it could batch the things and give me max one PR every X days instead of every time it happens
Guest73396 has joined #ipfs
master8787 has quit [Quit: Leaving]
dysbulic has quit [Quit: ChatZilla 0.9.92 [Firefox 41.0.2/20151016093648]]
<dignifiedquire>
victorbjelkholm: jbenet whyrusleeping any ideas what this error message means: “context deadline exceeded” when trying to do a request via the http api?
<victorbjelkholm>
dignifiedquire, my understanding is that it took too long time to find/download the file. Hit it a few times as well
<dignifiedquire>
daviddias: sounds like you are having the real fun here
<daviddias>
it could be related to the fact that we are getting different, almost chaotic, results when each of us tries the latest station app
<cryptix>
dignifiedquire: it's def. a timeout error
<dignifiedquire>
cryptix: thanks
<cryptix>
but wich part of the system... *shrugs*
<cryptix>
id love to go berserk on all the context error handling and add some actual context info..
Encrypt has joined #ipfs
null_rad- has quit [Excess Flood]
null_radix has joined #ipfs
Encrypt has quit [Quit: Eating time!]
<zignig>
cryptix: ;P
* zignig
has put the midget human units to bed and now has some time to talk.
<Teeed>
hm...
<zignig>
Teeed: ?
martinkl_ has joined #ipfs
arpu has joined #ipfs
OutBackDingo has quit [Quit: Leaving]
OutBackDingo has joined #ipfs
OutBackDingo has quit [Client Quit]
nessence has quit [Ping timeout: 260 seconds]
OutBackDingo has joined #ipfs
Joltidly has joined #ipfs
vanila has joined #ipfs
<victorbjelkholm>
just to make sure, if I have two computers connected to the same network, running ipfs, but the network has no internet connection, I can still cat/get content from one to the other, right now, no?
<victorbjelkholm>
Second question, if the network has a internet connection, it still prefers to use the network without routing through the external network, right?
<achin>
on the first question: yes you can. you might have to manually connect them, though
<cryptix>
victorbjelkholm: i wouldnt bet on 2) beeing true but just by latency the local node should be able to provide much faster
<zignig>
I'm working on a openstreetmap <-> ipfs proxy.
<zignig>
sort of working.
<cryptix>
yey
arpu has quit [Quit: Ex-Chat]
fingertoe has joined #ipfs
ipfsfan has joined #ipfs
<ipfsfan>
Hi, I was playing with the python API client, and couldn't seem to be able to publish to /ipns/hash-of-pubkey/arbritary-string, which I have seen mentioned. Is this implemented yet?
<ipfsfan>
Publishing to /ipns/hash-of-pubkey worked fine
<demize>
You publish a directory with a file called 'arbitrary-string'
<ipfsfan>
Thank you, so whenever I publish, I need to republish the whole directory?
<ipfsfan>
And does the arbritary-string file just contain the IFPS hash I want to point to?
<stoopkid>
the arbitrary-string file is an arbitrary file called arbitrary-string
edcryptickiller has joined #ipfs
edcryptickiller has left #ipfs [#ipfs]
<ipfsfan>
Thanks, it's clicked now
fingertoe has quit [Ping timeout: 246 seconds]
<stoopkid>
and afaik yea you have to republish the whole directory, and if this is not the case then i'd like to know, heh
<victorbjelkholm>
if the file changes, the hash of that file changes. And if a hash of a file changes, you'll need to change the links of the directory, which leads to the directory having a new hash. So yeah...
<victorbjelkholm>
but still, the other files doesn't change so it's not that big of a deal
<ipfsfan>
If you had a lot of files, wouldn't it take a long time to republish the directory?
<ipfsfan>
Or would the idea be to use multiple keypairs, each for a certain purpose?
sonatagreen has joined #ipfs
<ipfsfan>
Does the API allow the client to supply some hashes, then make a "virtual" directory, skipping the need to read all of the data (and the need for a real on disk directory)?
martinkl_ has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<cryptix>
ipfsfan: yup - look into 'ipfs object get' put and patch
<cryptix>
im not sure how feature complete python-ipfs-api is, though
<mungojelly>
oh i didn't know there was a python api at all, that's nice
<whyrusleeping>
that happens for everything on our api so that the trailers can work correctly
<whyrusleeping>
;w
Matoro has quit [Ping timeout: 260 seconds]
<whyrusleeping>
hmm... vim this is not
<dignifiedquire>
same error happens if I disable chunking
<dignifiedquire>
so I do not think that’s related
sonatagreen has quit [Remote host closed the connection]
<ipfsbot>
[go-ipfs] whyrusleeping force-pushed fix/log-hang from f21fe5c to 3727f48: http://git.io/vlJ3U
<ipfsbot>
go-ipfs/fix/log-hang 3727f48 Jeromy: fix log hanging issue, and implement close-notify for commands...
<dignifiedquire>
jbenet: if we can get that request to work, it would be very awesome, as that would mean we can drop all the vinyl stuff from the js-api :)
<whyrusleeping>
dignifiedquire: i can take a hack at it in like, an hour
<whyrusleeping>
maybe soon
<whyrusleeping>
er
<Nukien>
noob question - is the gateway (gateway.ipfs.io) simply another node in the mesh or is it special somehow ?
<dignifiedquire>
that’d be awesome :)
itradarcz8787 has quit [Read error: Connection reset by peer]
<Nukien>
Considering ipfs not connected to ipfs.io
<dignifiedquire>
I’m gonna push my code to github so if you want you can reproduce it
itradarcz8787 has joined #ipfs
<dignifiedquire>
whyrusleeping: code is in this PR if you want to test it yourself, just need to run “npm test”
<whyrusleeping>
Nukien: its kinda special in that the 'ipfs.io' domain gets round robined to a set of machines
<whyrusleeping>
but its really just like, ten different ipfs nodes
<Nukien>
OK so a private mesh would just be a bunch of nodes with perhaps a similar RR for some dns name. Those nodes would go in the bootstrap list ?
<jbenet>
dignifiedquire: the vinyl stuff is just a representation of a file-- we'll still need something like it (a js object to represent a file before reading it all in). the choice of vinyl is because vinyl-fs has a large ecosystem of tooling around them, but happy to switch it if there's something better.
<jbenet>
also, this makes sure to (a) read all the local filesystem first to make a list of all the files upfront, instead of later, as things may change. (b) has proper streaming, with backpressure.
<ipfsbot>
[go-ipfs] mildred opened pull request #1916: Remove usage of merkledag.Link.Node pointer outside of merkledag (dev0.4.0...ipld) http://git.io/vlmP0
<ipfsbot>
[go-ipfs] mildred closed pull request #1902: Remove usage of merkledag.Link.Node pointer outside of merkledag (master...ipld) http://git.io/vWKrM
tilgovi has quit [Ping timeout: 240 seconds]
<victorbjelkholm>
ok, as a golang beginner, I want to use ipfs from go. Should i use the unoffical bindings from go-ipfs-api or just integrate with go-ipfs directly?
fingertoe has joined #ipfs
<whyrusleeping>
victorbjelkholm: what do you want to build?
<victorbjelkholm>
whyrusleeping, gonna experiment with rewritting openipfs to golang instead of using node
<whyrusleeping>
oh, you can probably just use the api
<whyrusleeping>
using go-ipfs directly is generally for when you want to network protocol things, or manipulate internals a little more closely
<victorbjelkholm>
yeah, wanna do whatever is cleanest. Is there anything similar to js-ipfs-api for go? So I don't have write my own wrapper
<whyrusleeping>
like, if you wanted to write a p2p chat app, you would probably use go-ipfs directly
<whyrusleeping>
go-ipfs-api
anticore has joined #ipfs
<whyrusleeping>
github.com/ipfs/go-ipfs-api
<cow_2001>
ipfs daemon won't shut down for a loooong while after ctrl-c
wscott has left #ipfs ["Leaving"]
<cow_2001>
should i send a second signal?
<whyrusleeping>
cow_2001: yeah, if it doesnt close in a second or two, give it the second ctrl-c
<whyrusleeping>
all the important stuff is done almost immediately
<victorbjelkholm>
whyrusleeping, cool, that's the unofficial bindings I were talking about. Thanks :)
<whyrusleeping>
it hangs on weird network stuff that makes it really hard to diagnose or reproduce
<whyrusleeping>
victorbjelkholm: wooo!
<cow_2001>
ding :\
<jbenet>
the go-ipfs-api is really bad compared to js-ipfs-api
<victorbjelkholm>
jbenet, hah, tell me about it :p Missing like a lot! Probably better off going with js-ipfs-api after all
<victorbjelkholm>
my only gripe is that it would be easier to distribute a binary than the whole project + node runtime to be able to run the client
<jbenet>
victorbjelkholm: yeah sorry, very incomplete, almost no docs, few tests. we'll get on it as there's time. if others want to contribute to go land, this would be a good place.
<jbenet>
victorbjelkholm: indeed. deploying go is really nice
<victorbjelkholm>
jbenet, no need to be sorry! rather see an awesome go-ipfs project than awesome go-ipfs-api project, if you know what I mean
<jbenet>
:) indeed.
amstocker has quit [Ping timeout: 265 seconds]
<dignifiedquire>
jbenet: I will write something up tomorrow why I think that it would be very benefical for js-api to pull out the vinyl deps, too tired to explain anything in a useful way today
<whyrusleeping>
victorbjelkholm: yeah, the go-ipfs-api has mostly been 'i need this call, lets throw it in there'
<dignifiedquire>
jbenet: (the main points are code simplification, and better compat between node and browser version)
voxelot has quit [Ping timeout: 260 seconds]
dignifiedquire has quit [Quit: dignifiedquire]
sonatagreen has joined #ipfs
cryptotec has quit [Remote host closed the connection]
SDr_ has joined #ipfs
<chriscool>
Hi everyone!
<victorbjelkholm>
hello chriscool !
SDr has quit [Ping timeout: 250 seconds]
f[x] has joined #ipfs
<whyrusleeping>
chriscool: hey!
ygrek_ has quit [Ping timeout: 246 seconds]
<chriscool>
Tomorrow I will fly to the San Francisco bay area
<ipfsbot>
go-ipfs/wip/NodeTree 77735b5 Mildred Ki'Lya: Remove usage of merkledag.Link.Node pointer outside of merkledag...
<ipfsbot>
[go-ipfs] jbenet pushed 15 new commits to master: http://git.io/vlYFp
<ipfsbot>
go-ipfs/master 97a80e9 Christian Couder: Add first version of sharness_test_coverage_helper.sh...
<ipfsbot>
go-ipfs/master 741c0e5 Christian Couder: coverage_helper: remove more cruft...
<ipfsbot>
go-ipfs/master 5abf294 Christian Couder: coverage_helper: remove .ipfs...
<ipfsbot>
[go-ipfs] jbenet pushed 1 new commit to master: http://git.io/vlYbw
<ipfsbot>
go-ipfs/master 933b973 Juan Benet: Merge pull request #1912 from ipfs/t0220-bitswap-stat...
<jbenet>
i love tests. thanks chriscool
<ipfsbot>
[go-ipfs] jbenet pushed 1 new commit to master: http://git.io/vlYbK
<ipfsbot>
go-ipfs/master a93dbf6 Juan Benet: Merge pull request #1913 from ipfs/test-ipfs-config...
cryptotec has joined #ipfs
<jbenet>
whyrusleeping: shipped 0.3.9 yet? is it waiting on the logging stuff? (reminder that right now people cannot use the webui nor much of the api).
NeoTeo has joined #ipfs
<vanila>
windows binaries for IPFS?
simonv3 has joined #ipfs
bedeho has joined #ipfs
<whyrusleeping>
jbenet: that logging stuff makes the webui usable
<whyrusleeping>
sooo, yeah
<whyrusleeping>
waiting on that
<whyrusleeping>
and people can use the api just fine, not sure what youre talking about there
<jbenet>
whyrusleeping: what is that waiting on next?
<jbenet>
whyrusleeping: no, they cannot, diasdavid has been pushing for 0.3.9 for this reason alone. remember that merging 0.3.8 broke it without ensuring it was upgraded before releasing https://github.com/ipfs/js-ipfs-api/issues/87
<jbenet>
i believe the logging bug in the webui has been happening for a while. obviously we want it to be fixed, but remember that on monday we promised to ship out 0.3.9 on tuesday and it's now friday. we should really ship out 0.3.9 asap. let's not delay another day for it. if it makes it in, great, else it can go later. every day that passes is a day with a
<jbenet>
broken interface and users who try things from our docs that do not work.
rendar has quit [Ping timeout: 250 seconds]
<jbenet>
cc diasdavid o/ -- is the js-ipfs-api update ready to ship once 0.3.9 lands? (or is there one that needs to be there? dont recall)
<whyrusleeping>
jbenet: the daemon will actually hang and nothing will work AT ALL if someone uses the webui
<jbenet>
whyrusleeping: _ANY_ of the webui?
<whyrusleeping>
like, yeah. we wanted to get this out on tuesday, nothing got reviewed until wednesday.
f[x] has joined #ipfs
<whyrusleeping>
and yeah, the webui opens up the 'logs' endpoint when you open it, as far as i can tell
<whyrusleeping>
and it doing that screws up the loggers locks
nessence has joined #ipfs
<whyrusleeping>
which deadlocks any goroutine trying to call 'eventlog'
<jbenet>
whyrusleeping we then added other PRs because we deemed important, which is of course fine, but not good to do for a whole week. right now i'm not certain 0.3.9 will ship today.
<jbenet>
whyrusleeping "the webui opens up the 'logs' endpoint when you open it" sounds like a) bug in the webui that shouldn't happen, and b) not something new, becuase the webui and logs have been this way for a long time and it worked just fine.
<jbenet>
whyrusleeping: it may be that the deadlock was introduced recently, because it really was not a problem 3+ weeks ago.
<jbenet>
(probably wasted resources but did not deadlock)
<revolve>
it's an idea i've had since about 2011 that's just starting to work
Matoro has joined #ipfs
<whyrusleeping>
jbenet: we can disable the webui, or disable the logs endpoint and ship 0.3.9 if you want
<jbenet>
revolve: i can't run a py program atm. have screenshots?
elima_ has joined #ipfs
<jbenet>
whyrusleeping: it's up to you-- whatever you think will get done today. if you cant make the log changes today, then let's disable the log endpoint in the webui until it's fixed. it would be useful to figure out _why_ it started to become a problem now so we don't run into this sort of problems in the future (i.e. this is was introduced and means a lack of
<jbenet>
testing. if we need to run phantomjs tests against go-ipfs release candidates maybe we should)
<revolve>
haven't got any of the current frontend at the moment though
<jbenet>
what do you mean by collaboration then?
<revolve>
in that screenshot a copy is being told an instance bound to :8080 has news.ycombinator.com/ and then retrieves it from that instance and serves it
<jbenet>
how do you ensure integrity of the original content?
e-lima has quit [Ping timeout: 250 seconds]
<revolve>
everything's rendered in an iframe that can be toggled as contenteditable. on edits it sends whatever part of the dom is changing over the wire where other connected users reintegrate the changes
<whyrusleeping>
jbenet: i think phantomjs tests would be good. I beleive the issue was a result of the timeout on the http handlers context being removed
<whyrusleeping>
before, things would deadlock, but eventually resume when that context for the api request timed out
<whyrusleeping>
but that was removed at some point (i think?)
<revolve>
for the p2p side it's just checking the hash of the content is what it was intially reported to be
<jbenet>
revolve: if i can edit a site, then it's no longer the original site. which is fine for collab documents, if you can track version history with authentication, but not so good for websites per se
<revolve>
on the client side there's just a proof of concept; there's no UX to speak of there at the moment.
<jbenet>
revolve: looks cool :) -- consider doing this over ipfs-- we have all the p2p stuff worked out and you get the primitives you need for integrity (everything is merkle tree based, signed auth of mutable content, etc)
<revolve>
yeah and there's the awesome issue if spinning a bazillion AWS instances to report your edit of google.com/ as a good copy
<revolve>
excellent
<revolve>
well I'm glad you dig it. I think I'm going to continue having a go at solving these problems in terms of this project for the time being though
<jbenet>
whyrusleeping: "the timeout on the http handlers context being removed" have a link to a diff?
<ion>
I’m happy to see Go doesn’t suffer from the Billion Dollar Mistake and have nil inhabit every type. Now to figure out how to tell it I do want this one struct field to be nullable.
<whyrusleeping>
ion 'a *Type'
<ion>
whyrusleeping: Thanks. I tried that but then i wasn’t able to get a pointer to a constant with &namesys.ImmutableTTL.
* whyrusleeping
thinks ion is doing something really weird
<jbenet>
whyrusleeping: this impl looks very complicated to me. what are the actual constraints here?
Guest73396 has quit [Client Quit]
<whyrusleeping>
this is what you asked for in the CR earlier
<whyrusleeping>
you wanted each writer to be buffered, and to not block the daemon
<whyrusleeping>
and to not rely on a timeout
<codehero>
victorbjelkholm: nice. a "use local daemon" button
<victorbjelkholm>
codehero, yeah! With that feature + the application now working offline, it will work without internet access, as long as you're connected to the same network :)
<victorbjelkholm>
Now if we just had a chat to send the actual link the same way
<codehero>
however it has a few... problems
<codehero>
like when i press save, nothing happens
Guest73396 has joined #ipfs
<codehero>
and i can't deselect the "use local daemon" button
<victorbjelkholm>
codehero, I could be more clear in the UI but you can only save once the text is change. And the url does change! You'll need to have the daemon running and CORS activated
<victorbjelkholm>
you should be able to deselect the "use local daemon" though. Which browser+os ?
<victorbjelkholm>
you get any errors in the console?
<codehero>
archlinux + firefox nightly
<codehero>
well
<codehero>
"n is null"
<codehero>
doesn't sound good
<codehero>
there are so many errors
<victorbjelkholm>
huh, bleeding edge huh. You're loading it from ipfsbin.xyz ?
<victorbjelkholm>
Ah, that is not working because the daemon API doesn't work via HTTPS and ipfs.io doesn't work via http
<codehero>
oh
<victorbjelkholm>
browser doesn't allow request from https -> http, so if you want to use local daemon, you would have to run it locally or via ipfsbin.xyz
akkad has quit [Excess Flood]
<codehero>
okay. it works with ipfsbin.xyz
akkad has joined #ipfs
<codehero>
but it doesn't work locally
<whyrusleeping>
jbenet: pushed updates
<victorbjelkholm>
codehero, errors?
<victorbjelkholm>
probably you didn't set API_ORIGIN="*", no?
Guest73396 has quit [Ping timeout: 250 seconds]
ipfs_intern has joined #ipfs
<codehero>
victorbjelkholm: same errors "n is null"
<codehero>
and Network error
<codehero>
oh
<codehero>
there we go
<codehero>
Cross-Origin Request Blocked: The Same Origin Policy disallows reading the remote resource at http://localhost:5001/api/v0/add?stream-channels=true. (Reason: CORS header 'Access-Control-Allow-Origin' missing).
<victorbjelkholm>
so when you run daemon, you are not running it like this: "API_ORIGIN="*" ipfs daemon"
<victorbjelkholm>
which is needed to allow CORS
<codehero>
what's CORS?
<victorbjelkholm>
security for domains not being able to make requests to other domains. But since you want to use localhost:5001 from ipfsbin.xyz, you have to allow it
jamie_k has joined #ipfs
voxelot has quit []
<codehero>
ah
nessence has quit [Remote host closed the connection]
ipfs_intern has quit [Quit: Page closed]
nessence has joined #ipfs
sonatagreen has joined #ipfs
tilgovi has joined #ipfs
<cryptix>
codehero victorbjelkholm: API_ORIGIN is deprecated, see ipfs daemon --help
simonv3 has quit [Quit: Connection closed for inactivity]
<victorbjelkholm>
cryptix, oh, good to know. Way easier than writing those three commands though. Also, editing the config manually would also be easier than that
<cryptix>
well.. just fyi :)
cryptotec has joined #ipfs
Matoro has joined #ipfs
<codehero>
cryptix: ah. thx
<jbenet>
whyrusleeping: seems roughly correct to me, but im not sure. lost in channels. seems overly complex. i suspect something cleaner exists with moving the concurrency to the children, and using locks in the parent. maybe ask cryptix for his opinion
<jbenet>
prob will want to test the hell out of it before landing it