#dat

ogd: pr a close method that closes all the writers

mafintosh: remote-update question, if i have a hyperdrive can i do drive.metadata.on('remote-update', cb) to ensure the hyperdrive has finished a bitfield exchange?

ogd: yup

ogd: and you can send a pr that forwards the remote-update event to the drive instance if you want

mafintosh: ok have prototype dat-npm working

mafintosh: are you at a pc to try it out real quick?

laptop*

yuh

ogd: what do i run ?

mafintosh: clone https://github.com/mafintosh/dat-npm and run get.js

mafintosh: to see if that part works

mafintosh: should print a bunch of json and then eventually process should exit

ogd: installing deps

ogd: prints a bunch of json

mafintosh: cool

mafintosh: now run cache-get, it should print a stat object

ogd: it does1

mafintosh: ok so

mafintosh: how change cache-get to another tarball you havvent requested yet

mafintosh: e.g. hyperdrive 9.5.1

mafintosh: then open http://harvest.publicbits.org:8080/cache/reques...

mafintosh: but with the module/version you wanna download

mafintosh: doing a GET on that api writes the tarball to the tarballs hyperdrive and then when thats done returns 200

mafintosh: so after you GET the above URL you should be able to run cache-get again and it should get the stat object for the tarballs you just requested

ogd: one sec

ogd: you should try and stat it before remote-update as well

ogd: if it is there already, fine - otherwise wait for remote-update

ogd: hmm, http://harvest.publicbits.org:8080/cache/hyperc... didn't seemm to work

mafintosh: try deleting the .db folders locally and running again

ogd: same

ogd: can you download the o/ module

mafintosh: on my phone now

mafintosh: hmm the tarballs are definitely getting written to the hyperdrive on the server, i just double checked...

mafintosh: o/

mafintosh: maybe the remote-update is firing from a peer who has an old copy? anyway to get an event for when the writer updates?

ogd: keep listening for remote-update and keep trying

mafintosh: oo i get a lot of remote-update events

mafintosh: but seems to work

ogd: many peers :)

mafintosh: is there an event for when the content of the hyperdrive is updated?

mafintosh: i guess i could check the latest seq number...

ogd: on(append)

mafintosh: ah i see though now i should just wait till one of the remote-updates makes the stat comes back true

ogd: yea what i meant before

ogd: and check it right away also

mafintosh: yea nice, works great

mafintosh: pushed fix to master

ogd: we can make a wait: true flag that does this behind the scenes

mafintosh: oh yea

mafintosh: can i iterate every key in a hyperdb somehow

ogd: not yet, soon :)

ogd: works great now

hey all—haven't shared this too widely yet, but thought to drop the link here. it’s sort of a dat CMS using choo as a little front-end framework: https://enoki.site/

adding a little “publish” button which syncs with dat in the panel is one of the next big features on the roadmap

ooooo

anyway, just added the information about the panel to the site, and some writing to the log

ogd: latest hyperdb has wait: true on .get

mafintosh: hah nice

ogd: so db.get('hypercore', {wait: true}, cb) does all the magic for you

kinda nice

mafintosh: im doing it w/ hyperdrive in this case but nice to have nonetheless

ogd: we should have the same there going forward

jondashkyle: how do i login to panel?

ogd: you can run the panel by cloning the starter kit, then running `enoki panel`. it's an early preview, but there will be a hosted version, too. also working with yosh right now to get the local development server running on his updated bankai, which will smooth out some of that

cool

jondashkyle: wow panel is super cool

anyone ever compared http://xanadu.com/ to dat. There seem to be some overlapping goals/solutions, especially with Beaker

pfrazee: ^

creationix: I have made that comparison actually

of course xanadu is the ultimate omega project (using domanic's term)

also they seem really bitter about the tech that won out

oh yeah. You ever read that wired article about it?

Ya that's quite a project

https://www.wired.com/1995/06/xanadu/ this is a great read

^ should be required reading for a CS major

anyone see that vid ted published? https://www.youtube.com/watch?v=mmfjM-SGlGs

(have gotten distracted and only made it maybe 15 min in so far, ha)

someday I'm going to make a code school called "CS School of Hard Knocks and Dead Locks" and that'll be part of the curriculum

lol

jondashkyle: yeah he has a lot of videos like that. He's kind of a tragic figure

yeah…

pfrazee: only way to get it in is through port knocking and then passing the right key?

the school that is, to get into the school

hahaha

"computer world is a nightmare, honky tonk prison, gone wrong in so many ways..." - Ted

I mean, Ted has very little appreciation for engineering or business, and (arguably) not enough appreciation for real UX concerns

Even when they had a million-a-year development budget, they spent all day having fun instead of building actual products

his biggest grievance with the Web is that it used one-directional links, and he calls them "jump links" because they change the page, when he wanted those links to open up the target doc side-by-side and visually show the link between the documents

yeah. You kind of have to take him lightly. He's a man who had a lot of good ideas but wasnt very good at execution, and blamed all his failures on other people

jondashkyle: ted is a cool guy, seems like he gets caught up in in architectural ideas rather than implementation

bret: yeah, which is to say we need people like that too. a lebbeus woods figure of computer science, altho lebbeus didn’t carry the same type of bitterness with him.

bret: seems like his own social inabilities had a large role in the friction

ya'll treating philosophers like they're supposed to be engineers :)

haha! totally.

yeah exactly he perceivably caries a lot of bitterness, exactly.

ogd: problem is that ted, by his own admission, wanted to be steve jobs

and I think you gotta be careful about ted because his judgment, even from a philosophical standpoint, is from that bitter perspective

doing some napkin math on npm, they dont publish tarball sizes in the metadata. so im doing a HEAD request on all the tarballs, there are 3.5 million tarballs.

the first 60k tarballs were 24gb

1.68 TB?

i got 1.4tb

oh yep, 1.4

so not bad

maybe i should just download them all

i have 64tb at my house :)

ogd: two questions. 1) what are those HDs like? do you have, like, 5 NASes? did it cost a lot? 2) how long till your ISP comes a knockin?

ISP has no bandwidth limits

gotcha cool

ogd: i'll mirror npm as well

Im pretty impressed yall are tackling that

been half-tried like 10 times already in the p2p tech community, so yall may finally pull it off

damn, so npm is only around 2-4 TB?

1-2

damn 1-2 TB?