Areas

Memory is laid out in areas — slash-separated paths:

work/antarius/suppliers
work/antarius/tenders
personal/health

An area is given on a write and narrows a search. Nesting runs downward: a question about work sees work/antarius and everything under it.

Why bother

Without areas, personal and work sit mixed together and a question about a supplier turns up a note about a doctor. With them you can say:

— Only search work. — Put this in personal.

Naming them

An area is an area of work, not a thing.

Good: work/antarius/suppliers, infra/beelink, personal/property.

Bad: an area per company or per product. Everything about one company is already reachable by asking about it — a company is an entity, not an area. And the contradiction check works inside an area: the same statement written into two sibling areas never collides, and you end up with two copies.

Seeing what exists

— Show me the areas of memory and what is in them.

You get the tree with the number of events, facts and todos in each.

Describing one

— Describe the area work/antarius: medical equipment, supply and tenders.

A description helps you and the assistant choose where to write. Put in it what does not change: the kind of work, not the current participants and not the current numbers.

Renaming and merging

Areas are renamed or merged from the GUI or the server's command line, never through the assistant. The reason is simple: after such a move somebody has to know where the memory went. A person who named the destination and confirmed does; an agent that did it in passing does not.

The operation is irreversible in the data: merged areas do not come apart again.

The root

The empty area is the root, and an ordinary search reads it as "everywhere". What lands in it is what was never given a place, and it is worth clearing out now and then:

— What is sitting in the root with no area?