# dat protocol Community communication

27 Feb 2020

## People

* Diego (@dpaez)
* Eric ([https://github.com/erangell](https://github.com/erangell))
* Mauve
* edrex
* micah
* [bret](https://github.com/bcomnes)
* Kevin Faaborg
* Calm Rad
* substack
* many others

## Meeting Notes

### Intro

- people introduction
- Eric: Working on mentoring sessions with Mauve
- Edrex: Want to make apps with Dat. Want the SDK to work well. Looking for direction for app developers to future-proof their apps and make it as effortless as possible. Having a way to do storage at the user level and make it transparent to the app
- Micah: Fascinated with Dat and dweb stuff before it was even called dweb. Thinking about how to use tech in times of uncertainty. Working at an education technology non-profit on offline learning software and content curation.
- mauve: updating dat-sdk (node updates). browser needs help (funding). working on a new project using electron and dat, making it easier to share files on local networks. new module: multi-hyperdrive. Also working on dat-dot project.
- bretchs: Recently been setting up a hyperdrive-dht server. Relatively simple, deploying it to traditional aws infra managed with terraform. Deploying the DHT server to an ec2 instance with a domain name. There's a gist about setting up your own instance. If you ahve your own bootstrap server and you set that up you can basically create your own private DHT for service or peer discovery. https://gist.github.com/mafintosh/ec498f6f1ec32bfb74b777137b6f7d55

- calm-rad: Working on a mobile Dat browser. Using react-native and a mobile web view. Want to split out into own library for people to integrate with. Hoping to publish code after cleaning it up. Gonna post a link on IRC.
- Kevin: Working on Dat through the Ara project: ara.one. Working on an electron app, creators can distribute their podcasts over p2p and people can pay for it using the Ara token which is going on it's first exhange at 11 UTC today. Been busy on the telegram chat for Ara, coding electron otherwise.
- Serapath: Working with Nina and Mauve and some others. It's a hosting service for hypercores. Any data structure that is built on top of hypercores can be hosted. There's all these projects built on Dat have different data strucutres, trying to figure out how to make it so a dat address you put in to get hosting will find all the hypercores. https://github.com/playproject-io/datdot-service/issues/7
- substack: Been working on Peermaps. Just published [eyros]() a DB that is distributed over hyperdrive. It'd be good to use the Dat rust imp, but it's good enough to use the JS one for now. Next thing is working on the ingest pipeline for planet OSM files. Then getting the weekly patch files. Other peopel are working on the graphics and other parts.

## Standard storage stuff
- edrex: It'd be nice if we could standardize storage between platforms. I don't wanna write apps that only work in Beaker Browser.
- Kevin: I also don't wanna just write apps that work in Beaker
- diego: The Dat SDK was going in the same direction to make it easier.
- edrex: Thinking if this could be something with a UI that the SDK ships, maybe a set of UI conventions that tell people how - there's a wild west where there's lots of data being dropped on the floor. Every developer has to figure out how data is persisted. Maybe we need a way to make browser storage to be more solid

- diego: julian asked around for links to UX. Will link to tweet: https://twitter.com/juliangruber/status/1232237411712864256
- Serapath: Planning on having an electron app for datdot where people can add archives. Also so they can view websites through a local server. People could add the website as an iframe and use postMessage to talk to it and connect to the dat world. Haven't written how it works yet, using a server there might be some CORS problems, but the iframe could help.
single pixel / postMessage
dat polyfill
- mauve: mentions their exp with a previous app and sharing data using iframes. mauve found some issues with different domains. webextensions seems the safest way to go. browser: this parent frame shouldn't have access to this child frame. SDK as fallback
- substack shared https://github.com/substack/hello-frame-rpc
- micah: Recently checked out GNUNet. They have a simple resource identifier scheme `ghunet://` and then the module. So `gnunet://somemodule/somehash`. For the longest time I didn't understand how protocol labs ecosystem works. But after checking it out it looks like they took a look at what gnunet was doing. Wondering if peopel checked out their multiaddress naming scheme looks like. It'd be interesting to see what it could look like if Dat wasn't a unified protocol any longer. Maybe it'd be good to open up the naming. Somethign that's not changing but is open to more use cases.
- e.g. if dat followed a gnunet style naming scheme we might see stuff like:

    dat://hypercore/1234abcd....
    dat://hyperdrive/1234abcd....
    dat://kappa-core/1234abcd....

- edrex: It'd be cool to adopt mutliaddress and a lot of protocols are doing that. I wonder how the Dat community deals with Not Invented Here syndrome. It's the case that there's a lot of similar tech on each side.
- mauve: about adopt multi-address, it involves lot of things. in the end it's a huge thing to adopt. something to consider. think about random-access family of modules. small and concrete set of modules. another interesting case/approach is hyperswarm. there is a different approach in the dat community vs ipfs.
- mauve: there are some protocol schemes in the dat comm. eg: cabal, cobox. Nothing stops apps/community to come up with their own. It's a diff story to have native browser support.
- edrex: standars protect people.
- mauve: agree with that. dat-consortium is an effort to bring conventions/standardisation. but since there is no central authority the approach is a bit different.
- calm rad: Thoughts on if a browser were to have the ability for users to add more protocols in settings or something, where a protocol name is affiliated with any necessary libraries that would be needed on top of a basic dat layer? Maybe that approach would allow for the freedom in naming schemes while minimizing exclusion?
- mauve: [libdweb](https://github.com/mozilla/libdweb) project from mozilla some time ago was about that.
- Micah: I like how the community is pretty anarchistic. People can do their own thing and they support each other.
- serapath: Would be interested to see what data structures exist over Hypercore. There's hyperdrive, kappa-core. Maybe there are more structures? There's no standard right now so we should figure out how to get all the hypercores for datdot.
- searapath: I think corestore is a good library for that.
- diego: Yeah, people are using corestore more, instead of managing hypercores themselves.

### Links
Eric's latest mentoring session: Dat Peers and Dat SDK
https://www.youtube.com/watch?v=7DMCmXpMQ74