Category: Uncategorized

Identifies, indexes, and manages music files, and exposes them for playback via the Music Server Protocol.

Difference from Concept: Music Player Daemon: the playback occurs on the client, not the server

Differences from Navidrome: no built-in web interface; supports editing of tags.

Differences from Subsonic API: supports tag editing; has a concept of permissions; mult-user; more efficient than XML (I hope).

Differences from Plex: Music only.

Jellyfin: architecture looks good, all open source, but it seems video-focused and like it would suck for managing music; also seems to be .net-based

Totally novel API ?

Requirements:

• Scanning and discovery
• Uploads
• Permissions / users and ownership
• Tag editing
• Playlist building, names, images, descriptions
• Reencoding / transcoding
• Desktop and mobile client support
• Low-bandwidth connection support
• Gracefully handle albums, multi-artist albums, singles
• Music only. No videos, no podcast-specific support.
• Offline mode

Coming back to local Linux music playback, I’m pretty disappointed with the state of things. Spotify’s UI feels nice in comparison, which says something, given how neglected it seems. Things largely feel like 2016 or so. Every music player seems to solve the same problem using a totally new codebase, with only low-level libraries shared between them (reading MP3 or FLAC files, for example).

I can’t be the first person to have this idea… but, do like neovim did for text editing, like the Language Server Protocol does for coding: split apart core music playback functionality in a UI-agnostic fashion, and let interfaces interact with it via a socket protocol or whatever the equivalent of that is nowadays. (D-bus? MessagePack? no idea)

Handle the file formats, indexing, tag editing, output interfaces, notification of events, maybe interaction with online music repositories (e.g. FunkWhale)

Like this document. But in the open. Allow for comments on my project ideas, refinement, collaboration, etc. to bring things to reality.

this blog

My stab at evolving a self-replicating virtual lifeform.

A system for transcribing and selectively exposing your old journals.

Option: ship your journals to us, we use ML models to automatically transcribe and integrate everything.

Allow people to select which pages / passages, which people, etc. they want to reveal, and to whom. Set minimum age (e.g. only release 20 year old entries or older)

Automatic named entity recognition; similar to what photo services do with photos

Q: How good is handwriting recognition getting?

Building off of Concept: Git Blame Through Time: a graphical user interface that combines the best of git gui, gitk, and VSCode’s GitLens, plus the “through time” aspect.

Cross-platform. (Windows, macOS, Linux). Keyboard shortcuts for everything. NO ELECTRON. Attractive.

View of a file where a timeline of the edits/blame points to the file locations affected

Allow highlighting of commits that are ancestors of a target branch, e.g. master or develop

For a given set of lines in a file, and a given destination hash, show the history of the lines before a certain point, and as they were modified on the way to the destination.

START -> … -> P -> … -> DEST

Visualize the way the lines change over time, forward and backward. When did they enter the codebase? When did they leave it, or are they still around?

Interactively answer similar questions to git bisect, git blame, etc.

Genetic programming of operating systems

Initialize many VMs on a network with a starting kernel that can clone itself with some error probability, given sufficient resources. The “environment” provides the resources by minting them cryptographically?

The hope is that evolving the ability to access more of its hardware resources will allow the VM to better compete with other VMs. By developing better network defenses and offense capabilities as well.

Ability to store data on disk as well.

Everything the VM provides is available if it can but learn to initialize it.

VMs must be extremely lightweight to allow many generations.

On what basis do they compete?

Maybe initialize with a very basic unikernel OS—network stack, filesystem, etc. and error-prone “reproduction” facility.

The “system” allows cloning by proof of work. The better the work, the more reliable and/or faster the copy.

The “system” also allows network connections, limited by proof of work. The better the work, the more reliable the connection? Or the faster?

interface WorkToDo {
  reliabilityNonce: string;
  speedNonce:string;
}
// The number of leading 0s on the hash indicates the value
interface WorkDone {
  reliabilityHash: string;
  speedHash: string;
}

CPU and memory and bandwidth are the fundamental constraints.

If they can deprive others of those resources through hacks, they can outcompete them.

What gets copied?

Rust code?

Raw machine code? <–

Synchronize across devices and browsers—the rules, and the usage data

More customizable rules:

• Total block on certain days, time limit on other days
• Maximum number of days in a week on which to access the site
• Arbitrary boolean expression builder

People pay for the sync service?

Revenue share with original author(s)?

Alternatively: Something running on the router-server that things like Pome could be running on???

Domain: pome.network (registered 20 Jun 2022)

Indie poetry publishing.

Use Lightning Network to allow charging per view, or for long-term licenses.

Also allow donating to and making recurring donations to poets.

Poets can operate independently, or as part of a collective.

Backed by a peer-to-peer network by which the texts, and information about the poets and collectives, are distributed.

Easy integration with Lightning Service Providers and local nodes would be key.

Allow display of prices in dollars (optionally) based on an external API that fetches conversion rates.

A desktop application with graphical user interface to start.

Mobile version could follow.

Picking the cross-platform UI approach would be key as well. Tauri? Something that gets us mobile and desktop in one codebase? Something that avoids Javascript??

See also prior “Passion economy platform for poetry” and “Platform for reviewing poetry, intellectual conversations, etc.”

“Accuse” works of copyright infringement, take it to random vote?