Brrr!

It's chilly in the house, raining outside.

I worked on some ideas for #:transmissions with Claude on phone in bed, only just got up, 13:30. Coffee was necessary.

I've set myself #:pivot:spooky for 2024-10-31, basically release a v1 alpha of everything. An artificial deadline to help me get things done.

Big problem now, I need a system for prioritizing tasks in the massive todo graph.

One of my key projects is just such a system : #:farelo. But it's not yet usable...

To start it bootstrapping me, I first need to capture project information.

Form-a-matic

#:fam code generator to make custom forms is Top Priority

I've got the offline node bits working ok : you provide a pseudo-Turtle template for a given vocab. This is converted to an intermediate (#:yars) JSON representation, next that is used to generate the HTML form.

After filling out the form and clicking a button, a Turtle representation of the fields should be presented. The form uses a script called extractor.js.

The issue right now is that I'm using a couple of libs written for node, made browser-friendly by #:tools:webpack. But debugging is really painful. I've not been in quite this situation before so I don't even really know what tooling is available. This might be a one-off, so I can't really justify the time needed to learn new kit and integrate it into my workflow.

But my best guess as where the bugs are most likely to be is :

my code (written with much help from #ai:claude)
code/lib integration for browser
libs

For 1, as it happens, I've barely touched one particular first-line approach : tests! Because of the node/browser lib issue, I reckon I'll try kind-of mocks.

Unit tests for methods will be needed. But to prepare Claude to make these, a quasi-integration test seems a good place to start.

Take a rendered HTML form file, fill in sample data as preset values, this is the mock. In node, by read the HTML file into a document model in #:tools:jsdom, check the operation of extractor.js with that. Prompt time.