# `Statifier.Document.Initial`
[🔗](https://github.com/riddler/statifier-ex/blob/v2.0.0/lib/statifier/document/initial.ex#L1)

An `<initial>` element: spec 3.6's way to name a compound state's default
entry point via a `<transition>` rather than an `initial` attribute.

This is a slot on its parent, `Statifier.Document.State.initial_element`,
not a `kind` on `Statifier.Document.State`.
`<initial>` has no `id` attribute at all (spec 3.6), so it fails the rule
the kind set follows ("a kind is a thing with an id a transition can
target"). v1 gave it a synthetic id
(`../statifier/lib/statifier/parser/scxml/element_builder.ex:522-525`)
purely to fit it into an id-bearing struct, and then had to exclude that
synthetic id by hand from unique-id checking and from the compiler's id
map. A slot has no id to exclude in the first place.

`transitions` is a **list**, not a single nilable `Transition.t()`, even
though the spec requires exactly one `<transition>` child. Both the
zero-transition case and the two-or-more case must survive lowering long
enough to be named: the validator's check 4 reports a `<state>` with both an
`initial` attribute and an `<initial>` element, and the same "hold the
wrong shape long enough to report it" principle applies to `<initial>`'s
own transition count. A `%Statifier.Document.State{kind: :history}`
already has this shape for free - its default transition is
`transitions: [Transition.t()]` with no special casing - so the list
representation costs nothing extra to reuse here.

The `<initial>` transition's executable content, which spec 3.6 permits,
is representable at no extra cost: `Transition.content` already holds it.

# `t`

```elixir
@type t() :: %Statifier.Document.Initial{
  location: Statifier.Parser.Location.t(),
  transitions: [Statifier.Document.Transition.t()]
}
```

---

*Consult [api-reference.md](api-reference.md) for complete listing*