#dat

jhand: that wouldn't surprise me

karissa: ah ok it's because utp-native require fails silently so there's no utp, and I had tcp turned off

there should be some catch probably for when there's neither tcp nor utp

pfrazee: ok think I see where its coming from. It is an importing bug of some sort. The downloader is expecting 4 blocks but there should only be 2.

pfrazee: if you clear the .dat folder between tests it works for now

jhand: ok good deal

blahah: aaaaaaaaa____ 95% sure i fixed the bug. Release coming up

mafintosh: nice, I'm just making a PR to make append-tree compatible witn windows

blahah: yea i saw. that should go in hyperdrive tho

blahah: as in, hyperdrive should convert windows path to unix style

like tar

ah ok

blahah: i still want that pr tho

but in hyperdrive

blahah, aaaaaaaaa____: okay, released. try re-install deps and let me know if it helps

mafintosh: you still want PR for append-tree?

and thanks RE fix

blahah: i want the pr but in hyperdrive. basically just run paths through that unixify module

OK

pfrazee, jhand try reinstalling deps. i fixed a bug that could be related to what you describe

mafintosh: kk

mafintosh: for the fix do I need to update both the origin and downloader?

blahah: yea

cool thx

it was one of those bugs where it is hard to believe it ever worked

i messed up a bisect

but dat is integrety checking everything so it would just fail and reconnect and then work lol

#failover

mafintosh: hah nice. I just call that "writing software"

mafintosh I'm still getting EBADF

mafintosh: do I need to delete the original .dat dir?

blahah: yea that one is not fixed haha

ah ok

:)

blahah: using dat-next-next right?

yeah

blahah: just use dat-next, me and jhand made it up to date

oh cool, but does it do --seed?

jhand: o/

that's all I want it for rn

the original source is working OK with sciencefair

blahah: ah ok, workaround for the dat-next-next issue is passing the key as well

but trying to add more seed nodes isn't working

that fixed that EBAD thing

huh, so it did!

:D

mafintosh: I reinstalled deps but I still get incomplete sync issue unless I delete the .dat for the test files on each run

pfrazee: ya think it's a mirror folder thing. I'll look after lunch.

jhand: cool

pfrazee: ok

aaaaaaaaa____: very curious to fix helped your issue

blahah: o/ same

mafintosh: eh?

blahah: if my fixed helped with your speed issue

for me just including the hash worked

oh, hmm I'm not sure - I'm getting good speed but takes a long time to establish the connection

I think same speed as before - 2MB/s, but it took over a minute for the dl to start

back in 30

blahah: but the speed is consistent now?

Gonna start improving network perf now that everything else seems stable

yeah speed seems solid for dat-next-next to dat-next-next - I'll check for electron consumer tomorrow

blahah: awesome

karissa: chatting with nokome and bmpvieira today and had a good idea for a grant proposal centered around building a software dependency management system using dat

nice

who might be pitch that to?

we*

ogd: check out https://github.com/noffle/peer-npm

mafintosh: it didn't :( but everything looks the same, so investigating if it was updated

ogd: whaaaaaa. that would be so cool

mafintosh: dat-storage@1.0.1 & hyperdrive@9.2.1 , still get a key and error (Error: EISDIR: illegal operation on a directory, read) when i `dat-next-next . `

aaaaaaaaa____: dat-next-next doesn't take any folder argument as far as I know - it's just "dat-next-next" within the folder you want to share

ya mafintosh and I really made that confusing with the two next things, sorry

will be great if dat-cli lands soon :)

mafintosh: so sparse + latest works now?

oooh same question \o

pfrazee: so I can't get that test failure to reproduce anymore, didn't change anything 😕

blahah: trying to figure out how to convert links in markdown source to a .bib file (arxiv told me i needed one of these, i dont know what it is). i just wanna have 1 markdown file

ogd oh haha - how are you currently making references?

blahah: i just have markdown links

ah, yeah arxiv requires fuller citation metadata

sure you wanna put it on arxiv?

blahah: i dont know lol everyone says its the cool preprint

blahah: just trying to be like the cool kids

haha

what's the paper?

blahah: https://github.com/datproject/docs/blob/master/...

here's a simple bib file https://github.com/blahah/scraperJSON-demo-pape...

ogd so if you want it to be like a full academic style preprint you should probably have some citations - kademlia for example probably has a paper

blahah: yea i figured, i just wanted to figure out how to do it in markdown

ogd you can do the citations in yaml frontmatter

and pandoc will convert them to bib

blahah: oh cool

blahah: got an example?

kinda buried in here http://pandoc.org/MANUAL.html#citations

blahah: how would the markdown link look to refer to the citation key or whatever?

just @key

blahah: is this normal pandoc or pandoc-citeproc? i was hoping to find a working github w/ all this cause the implicit knowledge required is pretty harsh

blahah: i figured im not the only person in the world to ever create a citation friendly pdf from markdown

citeproc is only needed for applying citation styles I think

in general @key cites inline, [@key] is a formatted braced citation

aaaaaaaaa____: yea just dat-next-next no key :)

suppress the author with -@ so like "Ogden asserted in his tweet (-@ogdentweet)"

jhand, blahah ya fixed the sparse issue as well. forgot to mention that

mafintosh: woop

aaaaaaaaa____: is the reconnect issue fixed?

ogd: I like to use bib files, or some external file of references, because you can export them easily from a ref manager

ogd: I think the conf papers I linked are about as simple as it gets, but you can move the citations into yaml frontmatter

mafintosh:

oops sorry about the second one

oh wait no I'm not, the second one is `dat-next-next .`

mafintosh: I was thinking about discovery-channel last night

heya

it may sound lame

but I think it could potentially run over http2

just on a custom port

blahah: yea --filter pandoc-citeproc is needed to make the yaml frontmatter thing render correctly, i just got it working

because, truth is - it needs to be multiplexed

and http2 is quite a good multiplexing solution

where you can pause individual streams

aaaaaaaaa____: that first one is weird. is that just trying to share a folder?

it will help to discovery-swarm too

since feed id will be known ahead of time

and will be sent by connecting party

mafintosh: yes

first one is as described here: https://github.com/mafintosh/dat-next-next/blob...

cd cool-dir && dat-next-next

second one blahah suggested, i.e. cd cool-dir && dat-next-next .

aaaaaaaaa____: there is no old dat in thete

ogd: ah sorry

?

no, it is a directory with a few files and no dat

got it

indutny: for browser support?

no, for everyone

see

right now you need separate discovery swarm for every feed

because it won't connect to the peer that it has already seen

this is a waste of resources

indutny: no you can reuse the connection to share multiple

we do that in hypercore

hyperdrive i mean

hm....