Registro Memorial

Live, with invite-based signup and Google Sign-In. It serves roughly 5,900 records across the yards currently loaded — a figure counted from the store rather than typed into a config, which is not a stylistic preference.

Why every count is a query

R21 has shipped a client site that advertised 8 cemeteries while serving 130, because the number came from a status enum somebody forgot to update rather than from the data. Nobody noticed, because the page rendered perfectly.

So on this build every displayed count is a count() over the store, and a figure that cannot be produced is shown as unavailable rather than estimated. That rule came out of this project and now applies across R21's work, including this site.

What the source data does not tell you

The public datasets behind a burial record are not as clean as they present. The clearest example: one federal source marks spouses interred in a veteran's plot as veterans themselves. The row is structurally valid, the field is populated, and nothing about it looks wrong — it is only wrong if you know what the field means.

That is the general shape of this whole project. The hard part is not ingesting public records; it is that a public record can be internally consistent and still assert something false about a real person, and the correction has to happen before publication rather than after somebody recognises their own family.

Search currently returns the first 200 matches. That is a known limit rather than a design decision, and it is on the list.