#dat

starting to look like dat = git meets bittorrent

pfraze: :) yeah!

pfraze: yea i'd accept that

pfraze: its on the website

lol

oh yeah lol

mafintosh: ok, I think somewhere in hyperdrive is not calling one of the write callbacks or end callbacks

because inside of bulk-write-stream, some of the callbacks never get called, and that is handled as the 3rd argument supplied from outside

substack: taking a look now

substack: okay found the bug

substack: its caused by a bug in osm but hyperdrive is silencing it for some reason

i'll pr you a fix and find out why it is being silenced

substack: https://github.com/digidem/osm-p2p-observations...

hypercore should throw an exception now as well

mafintosh: nice

mafintosh: I'm getting the same error in a completely different project actually

substack: try reinstalling hyperdrive and see if it throws

ok

there is another potential place this can happen. i'll add an exception there as well

mafintosh: on github I have the latest version already

same as npm, 6.0.0

substack: the fix was in hypercore :)

ok

i released it 5 mins ago

same problem

no errors

gah

substack: can i run the code somehow?

i'm curious why this wasn't fixed by my previous fix

mafintosh: https://github.com/substack/pkgdb

I'm running the example/cmd.js as `pkgdb` and then:

in a directory with an index.html file, I do:

rm -rf .pkgdb && pkgdb publish 1.0.0 && pkgdb versions

and I never get the 'finish' event from https://github.com/substack/pkgdb/blob/master/e...

ah i know this bug. gah

caused by .createArchive being called with the same key

its the same bug we had in osm earlier that i just sent you a workaround for because i was lazy

should probably just fix it

substack: fix coming up

so we should always be just using archive.key from now on when passing to createArchive?

karissa: no you don't need to change anything. i classify o/ as a bug

ok

sweet

:)

this is a good property you get by having people use your software. you find the actual bugs and not the ones you kinda make up

:)

having these little modules makes it so easy to find bugs, too. all open source should be like this

today i'm continuing to prepare a small demo for next week. going to have a .csv, some image files, a pdf, and some python and R scripts and load it into a hyperdrive with some simple styling to show off code + data + text as an archive in the browser

with eljee's help we're going to expand mafintosh's demo of viewing pages of the csv to demonstrate browsing even big data files in browser using hyperdrive

we could also use juliangruber's app to share the folder, then open it in a browser. i think it'd be a slick dmeo

demo

karissa: fuck yea! :)

karissa: i'm feeling pretty certain i'll have new dat out before your thing as well

karissa: +1!!!

mafintosh: awesome....!!!! we could then demo the share from CLI or the app

i only have 4 minutes but i think we could cover that all

even if the app isn't styled it'll still be a powerful demo

but i want to style it and eljee has some design chops

kriesse is going to try working through it too but she's just starting so no pressure

btw juliangruber talked to her today, she said it was helpful to chat with you earlier

juliangruber: you rock!

ogd: i wonder if eljee and I could get creds to dat.land -- i changed comps and I think maybe my new ssh key isn't on there, or that im using the wrong user account

substack: think i fixed it for good now. i'm gonna try out pkgdb

mafintosh: oh and a more advanced demo (not for next week probably) would be updating it and getting the history, even if they are just indexes (not pretty hashes) or having a history sidebar where you can see the list of hashes and click back in time

karissa: why thank you :)

mafintosh: the new hyperdrive api is super easy to grok and its fun to build on top of, best yet!

karissa: haha good to hear :)

mafintosh: i dont feel like i need an intermediate js api anymore

a missing minimist dependency is the most common bug i've found in substack projects

if you have any thoughts on this, too, the next step might be a hyperdrive-swarm module that does a single swarm for multiple archive github.com/karissa/hyperdrive-archive-swarm

karissa: yea thats a really good idea. thats too hard to do right now

kk. one day

i'll send some prs to your module soon

after shipping!

good yes of course

ill stop my post-coffee morning update/idea rant now

hah

karissa: its good stuff

substack: i get the finish event now in pkgdb

and then some other bug

substack: i think it is fixed for good now (bold statement)

mafintosh: i would love a streams class from you. so far my knowledge is limited to api docs and the stream adventure

clkao: cool awesome-dat repo

i need to use something else other than mafintosh's file-browser-widget for the demo because it reloads every time there's a file and its a bit clunky with react. might just fashion something quickly with yo or use shama's fs-explorer

karissa: +1 on that. i improved dat-desktop slightly by limiting the render function to 60 re-renders per second, but could still be a lot smoother

juliangruber: good idea. i might just do something simple.. working on it now. i think if we just use yo it should fix it :)

cause it'll do the diffing..

right?

karissa: i think simple and works is the way to go atm

yup

react comes with its own lifecycle and stuff, so not using that alone will speed things up

k ill swap it out in hyperdrive-ui

+1 on simple

just fyi im going to continue working on https://github.com/karissa/render-data unless someone tells me otherwise

also karissa: you want me to get .html rendering working?

eljee: i think it should already work well enough. the csv is more important

eljee: we can use packify to put assets into .html files

eljee: have you seen packify?

not yet...

what other functionality you want with csv then?

karissa: you want the ability to ask for specific rows maybe?

karissa: i need to also fix error handling and callbacks

eljee: i think a byte range could be good enough, knowing how big a row is could be a rabbit hole

honestly a preview with pages is probably fine

ok

for now

ok

eljee: do you have interest in working on the replacement for file-browser-widget?

karissa: sure

eljee: oh yeah you just said error handling. totally important

karissa: you have something in mind

eljee: just not using react

karissa: yeah, i also want to do a left sidebar with the tree and right-hand display window for design. do you think thats ok or did kreisse want to do something else?

eljee: she is off now. i think that would work well but i also like github's design

eljee: she is gonna start thinking about this stuff this weekend/monday

eljee: but getting something working is most important, so whatever is easiest/minimal

ok

eljee: do you want to pair on it?

honestly i work terribly in pairs

no worries i get that

karissa: HATE pairing. we could hack in the same room.

karissa: sorry just being honest about my limitations

lol :) ok. no that's good

A+ honesty

i work better together if we make a task list and each take things and then collaborate when throwing them into repo

okay want to screenhero real fast?

then we can parse things out

sure give me 5 mins tho

no worries

ok screenhero?

eljee: can you hear me

no

is your mic on?

i can hear you

one sec

let me refresh sh

clkao: eljee: ok for tuesday on dat.land https://github.com/datland/todo/issues

ogd: o/

suhweet