brainstorm ways to start using filecoin

phritz commented 6 years ago

Our goal for Q2 is to launch alpha, a persistently-running private internal testnet. This means we can have applications running on filecoin this quarter. We could even start building them before alpha has even launched. Let's talk about what some applications are and how they might work. Let's separate the question of what would give us good information from the question of who would build it.

Proposed Goals

primary goal: start shaking out the bugs and rounding off the sharp edges that clients are going to hit using filecoin to store and retrieve files
have actionable feedback for how to improve the filecoin client developer experience that we feed directly into our dev process in the form of stories. this could be api changes, pieces of libraries we need to supply, bugs to fix, performance problems to address, docs we need to write, whatever
minimize impact on dev team; isolate the impact to small number of people; this will be good practice for when we open up
move quickly, shoot for prototype-level quality

Proposed NON-goals

spend a lot of time on stuff for the use case that doesn't yield us actionable feedback for filecoin
take any of these over the finish line in the sense of producing a polished product; we're undertaking this effort to shake out the bugs at this point. Sure, some of these things might turn in to real products, but not at this point. We're still prototyping.
develop a killer app for filecoin (though we should consider spinning something like this out)

Process

Source ideas on this issue. Explore their feasibility and usefulness and then sort by decreasing usefulness / effort. Figure out a way to staff the ~2-3 highest priority efforts in Q2. Spin the projects up with a concrete work plan & process that involve a relatively small cost to a subset of the dev team. Organize the work so that it yields the best information (eg, select projects that cover a set of concerns and have at least some of these things be built by people who have not worked on filecoin or ipfs).

Deliverables

a prototype thingy checked into one of our repos or a report about why it was impossible / too hard to build; should maybe have a separation between its client library and the "app" built on top of it, that way we can compare client libraries and learn from what people do.
when the first major milestone is hit and something is working a brief report format TBD saying what worked well and what didn't
issues filed for problems that rise to the bar we set (see open questions below -- these could be anything from performance problems, to bugs, to missing documentation, etc)
very rough content for whatever documentation they found missing

Open questions

What if anything should we be doing to shake issues out of the miner side? Seems like we could have a similar effort for a couple of mining use cases (desktop use case, large deployment case, network attached storage case, etc)
What should the bar be for these people filing issues? My proposal is that they should file every single bug or problem they hit. Yes this is huge volume for us but might as well start collecting this info now, people are going to have to file it sooner or later. Will give us some material to start creating a backlog for contributors coming in, and will be good process for triage process. There is also the question of where to file issues; I say right into the main repo for good practice.

Existing ideas -- PLEASE PILE ON AND ADD MORE IDEAS, THIS IS A BRAINSTORM

filecoin backend for IPFS cluster: https://github.com/filecoin-project/notes/issues/26
folder sync: have a folder that syncs files placed into up to filecoin and down from filecoin into a new instance (obv to do this right is hard, just do it dumb and simple)
related to / instead of ^^: FUSE filesystem for my filecoin files
have gx store/install from filecoin
graphical wallet (@dignifiedquire say more here)
browser-based filecoin fetcher
ingest very large datasets

cc @whyrusleeping @dignifiedquire

laser commented 6 years ago

browser-based filecoin fetcher

Would you elaborate? This sounds interesting, and, I like browser-based things because I can show them to non-technical people.

phritz commented 6 years ago

Would you elaborate? This sounds interesting, and, I like browser-based things because I can show them to non-technical people.

Something like or prehaps even augmenting https://github.com/ipfs-shipyard/ipfs-companion#ipfs-companion

laser commented 6 years ago

That sounds excellent

porcuquine commented 6 years ago

Static website hosting. Yes, this would require some kind of web-sever gateway -- but it's a realistic use case so working through the UX/ergonomics might be valuable.

Configuration 1: standalone Go server that can be run on Digital Ocean / AWS / etc.

Configuration 2: same tool but run by a third party who serves FC content to associated domains. Associated how? IPNS? FCNS? Registered with provider (least good, but easy).

mishmosh commented 6 years ago

We will pick ~3 and get them running in parallel with combination of internal & external developers.

phritz commented 6 years ago

Enable https://medium.com/textileio/adding-the-next-million-peers-to-ipfs-76d356352d14 users to get paid for storage.

filecoin-project / venus