# `PhoenixKitKnowledgeBase.Errors`
[🔗](https://github.com/BeamLabEU/phoenix_kit_knowledge_base/blob/v0.1.0/lib/phoenix_kit_knowledge_base/errors.ex#L1)

The one place an error reason becomes a sentence a person reads.

Contexts return atoms and tuples (`{:error, :stale}`, `{:error,
{:forbidden_characters, ["#"]}}`); the LiveViews call `message/1` at the
flash boundary. Keeping the copy here means the vault page and the notes
panel say the same thing about the same failure, and translators have one
file to work from.

A reason nothing here knows is **logged, not shown**: an `inspect/1`-ed term
in a flash tells the person nothing and the developer less than the log line
does.

# `changeset_message`

```elixir
@spec changeset_message(Ecto.Changeset.t()) :: String.t()
```

Every error on a changeset, as one sentence.

Messages go through core's error translation — the same one `<.input>`
uses — so Ecto's own ("can't be blank") read the same here as on a form.

# `for_log`

```elixir
@spec for_log(term()) :: String.t()
```

An error reason, fit for a log line.

A changeset is reduced to its messages: it carries the note's whole content,
which has no business in a log. Anything else is inspected, bounded.

# `message`

```elixir
@spec message(term()) :: String.t()
```

A translated, user-facing sentence for an error reason.

---

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