like a Personal Server, but made up of mesh of functions running in various parts of the cloud. cf serverless
see old (2012-10-22) Udell Personal Cloud Series
2021 thread (which is bushy, hard to find all the strands): JSON, HTTP, ReST, WASM, composition, pipeline, Internet operating system, etc.
- on personal cloud vs personal computer
- on XMPP, matrix.org
- on OAuth vs alternatives (incl crypto)
- maybe WASM means you can run code locally, even on phone?
What problem am I trying to solve?
- compounding composition of software components
- for customization/personalization of each person's tools for thought: to clarify, I don't mean a closed monolith with personalization toggles/features, I mean a mashup of composable code-bits
- both open-source and paid (Making a living)
- available to all of a user's devices
Solution Candidate 1: Personal Cloud
- you have control/responsibility for almost everything
- everything runs in cloud
- you have "central" data store, roughly schema-less
- you run a collection of apps/functions that use that data-store
- many components are code you "load" and control
- other components are remote function-services: you pass a chunk of data, you get a chunk back
- some components are free, open source
- some are paid (pay-once or subscription) but trust you with the code to run
- some are paid-per-use functions
Solution Candidate 2: Own Your Own Data
- all you run/control is a cloud-based data-store plus ReST-engine
- all apps get/give data to that server, whether they run on your phone, laptop, or as a SaaS
Hosts to consider
- Glitch - runs on AWS, but doesn't have automatic access to AWS services
- Replit (which runs on GCP - but I suspect it can't access GCP stores, only the Replit-native options...)
- AWS - consumer-friendly enough?
- GCP - consumer-friendly enough?
- anyone else? Digital Ocean?
What Data-Store should this use? Probably a GraphDatabase... and where should it be hosted?
- Neo4j? AuraDB on GCP?
- JanusGraph on GCP with Google Kubernetes Engine (GKE) as the orchestration platform and Bigtable as the storage backend
- Amazon Neptune
- ah, postgresql!: (2022-08-17) Udell Postgres Everywhere, local-first
Is there value in pre-building item-types using W3 ontology standards?
- cf schema-less, OSAF Chandler, prototype-based, LinkedData (see Dorian Taylor stuff)
FederatedWiki might have more potential than you'd think: (2022-01-19) FederatedWiki 2022 activities.
Hmm, does the JVM have an important role? Pondering Clojure, etc.
- Neo4j is built in Java
The above is too handy-wavy, need to define/design a couple specific applications as examples...
Edited: | Tweet this! | Search Twitter for discussion