Lively Distractions

2024-09-28 Yesterday I was too busy doing things, head firing on way too many cylinders, so only wrote down reminders. I'll flesh them out a little now.

Tibico Research

Last month Oriana asked me to look after her jar of Tibicos aka water kefir. It's fascinating stuff, I've already started experimenting. It also made a nice test case for using Claude AI's Projects for research. I had only used them for software projects, for which they are great (with the help of repopack). I know virtually nothing about microbiology (though have been interested a while - I've got a usable microscope etc), so this topic offered a good virgin project.

I created the Project and set up a custom prompt. Then added the 'Edit' raw text of the Wikipedia page on Tibicos and that of Saccharomyces boulardii (a probiotic - that page is an interesting read), together with 4 papers linked from Wikipedia, to Project Knowledge. It worked really well, I immediately had a useful expert on hand.

I was curious how viable it might be to incorporate the probiotic critter into the existing symbiotic tibicos culture. Here's how the chat started :

Claude Tibicos research screenshot 1

By the 4th interaction we were at this :

Claude Tibicos research screenshot 1

Alife

Thinking about real (little) life reminded me of alife. I had to have a play with Claude. Not long ago danbri showed me things he'd been able to build.

Claudio

Farelo bits

write requirements first!

convert into prompt if appropriate

farelo metrics

sparql-store-lite

Many of my [hyperdata](https://hyperdata.it] projects require a SPARQL store. I'm in the process of setting up Fuseki on my server. It's my go-to for self-hosted, very capable and straightforward to set up. (Virtuoso and Ontotext are among other out-of-the-box alternatives, but I've also history of coding with Apache Jena libs on which Fuseki is built, so it's an obvious default).

Long-term, this time around I want to use LDAP-backed auth on my server which is taking a bit of time (I've already had to hack a bit of connector code for Apache Shiro. I used to know Java, eh...). I do have Fuseki running on my desktop, but while I was mulling this over I remembered that I'm meant to be writing some docs for Bergi's RDF-Ext. For which I need to get familiar with it. A little file-system backed SPARQL store running on nodejs would be a great little project, likely to be very useful for debugging #Transmissions.

So locally :

cd /home/danny/github-danny/hyperdata/packages
git submodule add https://github.com/danja/sparql-store-lite.git sparql-store-lite

Claude offered me a starting point, beginning :

import rdf from 'rdf-ext'
import N3 from 'n3'
import { SparqlParser } from 'sparqljs'
import { QueryEngine } from 'rdf-query-engine'
// ...

@rdfjs/parser-n3

nvm install node

npm install -g npm@latest

npm -v

webpack transmissions for in-browser

To interact with GitHub's API using Node.js, here are some suitable libraries ranked by activity, coverage, and popularity:

Octokit (https://github.com/octokit/octokit.js)
- Official GitHub REST API client for JavaScript
- Very active development, excellent coverage, and widely used
- Supports both REST and GraphQL APIs
@octokit/rest (https://github.com/octokit/rest.js)
- Part of the Octokit ecosystem, focused on REST API
- Active development, great coverage for REST API, popular
axios (https://github.com/axios/axios)
- General-purpose HTTP client, can be used with GitHub API
- Very active, widely used, but requires more manual setup for GitHub specifics

These libraries are well-maintained, offer good coverage, and are popular in the Node.js ecosystem for GitHub API interactions.