Alright, I'm dusting off an old domain that I bought and never used, and the maker space is rolling out an open media collective, discovery platform, and sharing space.

The time is now. is important.

I'll do it by myself if I have to, but shoot me a DM if you want to collaborate.

Think the Free Music Archive, or wikimedia commons or wherever kids are getting open license video game assets + a peertube instance, a curated blog, and a tip jar.

I have a rough draft around here somewhere.

Follow

I could use developers, reviewers, writers, and moderators, eventually.

In addition to people submitting CC licensed work, of course.

I'm going to start with a bare bones site and build up.

@ajroach42 Hmmmm, on the one hand I keep myself quite busy. On the other this is very much something I want to grow...

@alcinnz You do keep yourself busy.

I'll be happy to have you contribute or collaborate as much or as little as you'd like.

Let me get the basic org structure set up, and get a couple of folks up such that it's self perpetuating and we'll go from there.

@ajroach42 I can certainly serve as a reviewer, that's something I've already been doing!

@ajroach42 daygig's keeping my developer brain maxed out but I will absolutely write reviews of uploaded works.

@djsundog The more I think about it, the less I think we're going to need any original code.

I think I'm just going to use a static site backed by a gitea repo, with the assets themselves (music, stls, videos, etc.) hosted outside of the repo and linked in. (probably, assets will get hosted on a slow block storage device, uploaded over SFTP and mirrored on archive.org with a bot, but I'll figure that bit out in a few hours.)

@djsundog The site being stored in git means we can use it like a wiki, without the overhead of running a huge database.

In the event that we need any custom code, it'll just be the kind of devops-y kludges that I've built my career on.

Minimum viable platform.

@ajroach42 I wonder if we gain anything by using minio for object storage rather than a flat filesystem-based storage for the blobs

@djsundog I generally distrust additional complexity unless I can clearly understand what its for.

I don't understand what would be gained, beyond some flexibility and a fairly complex amount of overhead in the tech stack.

@ajroach42 I'm not sure either - replication and load balancing? probably achievable otherwise. file dedup can be done at the filesystem with these fancy new filesystems, so maybe it adds nothing? dunno haha

@djsundog

I've never had to think about filesystem performance on a personal project, and it is only now occurring to me that this kind of a massive media library might run in to some performance constraints on it's file system when dozens of users are trying to access files at the same time.

I solve this problem professionally on a regular basis, to the point that it doesn't even occur to me that it is a real problem.

@djsundog So, uhh, yeah some kind of object storage system is probably a good idea for this system when it outgrows running in a flat file system on an underspecc'd vps.

I will design with an eventual migration in mind.

@djsundog But for the v1, I'm focusing on keeping the tech stack as absolutely simple as possible, and I will adapt to usage patterns and constraints as they reveal themselves.

@ajroach42 @djsundog minio is backed by a standard filesystem plus some metadata, so you don't gain anything in regard to that. Maybe in a distributed setup (never tried)?

It gives you an S3 frontend to that filesystem though, which in turn enables access for a plethora of networked applications and libraries.

@ajroach42 @djsundog Minio doesn’t add that much complexity. It’s pretty simple to be quite honest. Clustering is it’s most complex feature, and overall Minio is fairly useless outside of single tenet object storage.

Gluster would be the next closest thing, and that is complex.

@djsundog @ajroach42 You gain, uh, lots of complexity. 😉 (And other things, if you need it, but if everything fits on a drive you have and doesn't need replication you can't do with rsync and a cron job... it's probably not necessary.)

@aschmitz @djsundog @ajroach42

It’s probably worth subdividing the files into a couple of layers of directory tree though the way git does with its objects.

@ajroach42 what kind of development? I've been wishing i had something better to do them play fortnite
Sign in to participate in the conversation
R E T R O  S O C I A L

A social network for the 19A0s.