#dat

oh yea ive seen a couple talks on secure multiparty computation its pretty interesting

(rukku) > ` karissa ` @rukku active development is being done on the rc2 branch, if you'd like to take a look!Got it. Thanks @karissa!

want to see something fun? https://www.youtube.com/watch?v=fdxBMpowoZg

if you want to play around some more, its in laurengarcia/netscape on branch `develop-hyperdrive-www`

i was hoping to do more with the server-side routing, that's why i picked up the hypercore -www bits and threw it into my repo that already had express ready to go on it

but it seems like there's a bug trying to read the archive key from the address (some kind of chunking error: net::ERR_INCOMPLETE_CHUNKED_ENCODING)

i tried to just run signalhub off my local but still no dice, same error. then i remembered that substack and mafintosh were going back and forth about an event not closing and figured that's probably it, so i just played around with the UI a bit.

have a good weekend everyone

btw if you want me to make a branch and PR out of this for hyperdrive-www, just let me know. wasn't sure how you guys want to break up the filesharing demo vs the registry.

eljee: the net::ERR_INCOMPLETE_CHUNKED_ENCODING errors aren't real errors, just noise

oh shoots. hm. do you know what its about then?

wondering if you've seen it

(the real bug that is -- the one preventing the read, or at least it seems)

if not no worries. it is reproducible by dragging/dropping files, letting the hash update and then hitting reload. seems like its trying to read the contents of the hash out of signalhub but not quite getting there? maybe i don't understand how it all fits together yet tho.

(by hash i mean archive key, sry)

eljee: looks cool! what do you think about making the whole body a drop zone by default?

eljee: WHOA

eljee: yea sometimes reload doesnt really work with the #hash links

eljee: because chrome doesn't reload the page, it just issues a onhashchange event

but there are workarounds for that

eljee: also the site looks great!

substack: this is a great idea

merged and released in 5.2.0

substack, eljee think i just fixed that ERR_... noise also

(andrewosh) *comes into dat chat to ask hyperdrive question, sees hyperdrive-index, leaves*

lol

our indexing approach is basically http://howfuckedismydatabase.com/nosql/

substack: do you need dex.ready or could you do a thunk internally so the first operation does idempotent ready?

eljee: i like the search-engine-esque minimal home page

ogd: when you write a hyperdrive-index, you provide your own storage mechanism, so you need to know when the indexes are caught up

substack: ohh i thought it was a different meaning of ready sorry

juliangruber: https://github.com/Level/memdown/pull/48

mafintosh: merged and publish pushed to github, unfortunatelly though i'm not yet an npm owner

pinging rod about this

juliangruber: cool, thanks

substack: finally got around to fixing this now, https://github.com/mafintosh/hypercore/issues/20

substack: what to you think the api should be? .hash(index, cb) ?

or .head(index, cb) is probably better since its the head of the feed at that point in time

how do i get the index?

index is just the number of blocks in the logs

-1

feed.blocks - 1 for latest

maybe it could just default to that

I think a method for getting the blocks would be good too

it sucks having to call .get() just so blocks is defined

yea agreed

thats on my list as well. its gonna be a range selection api

where you can give it a range of blocks to get and a callback that is called when all of them are downloaded

substack: api added, described it here, https://github.com/mafintosh/hypercore/issues/20

for the hash

cool!

i'll do a release now with all these new things.

technically its gonna be a major bump since this includes the final tweaks to the wire protocol

this is a cool project, https://github.com/andrewosh/container-share

hypercore 3.0.0 is out now

now i just need to upgrade hyperdrive to use that and then i'm ready to call it LTS

ogd mafintosh cool, im glad you like that demo. let me know if you want me to do any more with it. mostly just playing around trying to get to know it a little. am more than happy to put it somewhere more permanent than that rando branch.

juliangruber yeah i kept going back and forth re: drop zone stuff - using whole body or not. we could also make it the whole "overlay" but not the body. im sure the designer will have more input.

i will dig into hypercore/hypercore-index and signalhub more next week to understand more of what's happening underneath the apis

mafintosh what was causing that chunking ERR?

eljee: i think the whole body thing will more be for convenience, as i guess for initial explanation of how dat works a dropzone or file input are better as they reduce cognitive load

juliangruber yeah, i guess it can listen for the drop event on the body and open the panel accordingly

eljee: have you seen cloudup's interface? they do that pretty nicely

imo

juliangruber no, i'll check it out

eljee: so the signalhub uses a long lived http connection to do server-side events

eljee: but if no peers are discovered for 2mins the browser times out the connection

eljee: and prints that error. the client then just reconnects

eljee: the fix was just making the connection send a heartbeat once in a while

sounds almost romantical

juliangruber: i didnt know your band was so good https://www.youtube.com/watch?v=yF1IFRmFjl0