Journal 2025-05-31

This morning I was woken by Federico's tractor about 15 mins before my alarm. But that was fine, last night I prepared a flask of coffee. I did about 2 hours worth of dev with Claude before getting out of bed. Then went shopping, I've just fixed the Claude artifacts and I'm very happy. A very productive morning!

JSDoc RDF Plugin

It does what JSDoc does, with an output in RDF Turtle : jsdoc-rdf on GitHub. Something I've wanted for years but never got around to. It became more pressing now I've started integrating LLMs etc into my knowledgebase setup - in Semem (which still needs a bit of tweaking for my artificial midsummer break deadline).

OpenTelemetry in Transmissions

This seems a must-have for my pipeliney thing. Ok, every node in a transmission (= pipeline) is an implementation of a Processor. That interface is descended from ProcessorImpl which does the work of running the processors. This latter now has hooks into Open Telemetry. There is space for hooks in the code responsibility for the construction of pipelines, as well as in the API bits. But being lower demand, they are lower priority.

I haven't a clue yet how best to make use of this, I believe I'll need Prometheus and/or Grafana somewhere.

Filesystem watching for transmissions

Not unrelated to the telemetry thing. Right now, the #:Postcraft doc pipelines to render this blog process every single source markdown file. So I just added this fs watcher.

The idea is that this will be a long-lived background process. When a file in a given dir tree changes, this thing wakes up and does a HTTP POST to a given endpoint. For the #:postcraft use case, the intention is that there will be a #:transmissions mini-service that will receive the POST and run the markdown-to-sparqlstore, sparqlstore-to-html pipelines.

As yet untested, but the ability to trigger transmissions via HTTP is something I've had in mind all along. There is something approximating implementation, but I'm only now getting to the stage where I can start trying it out in practice.

Tip of the Day

I've currently got Windsurf hammering away at Semem. It was having diversions where it lost its thread somewhat, then repeated earlier approaches that failed... attention deficit, dude! But this seems to have helped :

I suggest you create a file WINDSURF.md and in it place note on strategies which work and those which don't areas of the codebase that likely need attention etc.

It's also a good read for me (same deficit). Also :

remember to update WINDSURF.md as you go along. Leave yourself a note in it to remind yourself about it.