A real page
The example data page
Loading the page's own export…
Unlike the walkthrough, nothing here is invented: this
is examples/spr.md from the repository, condensed from the working page
in njschooldata, and every
field below was read out of it by ergo.py export when the site was built.
What you are looking at is a render — the markdown file is the canonical artifact,
and it reads perfectly well without any of this.
The manifest
What this page is about
The registry
Issues
Each entry is one discrete thing a user of this data must know. Filter by what happens if you ignore it — the same closed vocabulary an agent filters on.
The decisions
Practices
Calls somebody made about what may legitimately be computed from this data, each naming the plausible wrong move it replaces. You find them by the question you are trying to answer rather than by what is broken.
In their words
Quoted from the publisher
What the agency itself says, with the URL and the date it was read there. Everything inside the quote is theirs; the note beside it is the page author's, and a reader can always tell which is which.
Prior work
Who else has parsed this
Pointers to implementations, documentation and articles about the same dataset — with the edition each one serves and whether it is still maintained. A reference records that something exists. It is not an endorsement.
The receipts
Validation
Dated records that the page's claims were checked against reality. A caveat nobody has tested is a hypothesis.
What this page costs to keep
Five issues, two practices, one quote and one reference, in a single file that sits beside the code ingesting the data. Most of it was written at the moment of discovery, because the workaround commit and the issue entry were the same change. That is the habit the whole format is trying to make cheap: an issue you found and did not write down is one somebody has to find again.
Read the raw markdown → · See the block reference → · Start your own →