Project layout

Nacara has no required directory names. Your site says where its files are, so the layout below is a convention, not a rule.

docs/
├── Docs.fsproj        the site is a project
├── Site.fs            its description, in F#
├── content/           markdown, read by a collection
│   ├── index.md
│   └── guide/
│       └── getting-started.md
├── static/            copied verbatim into the output
└── output/            generated - do not commit

The site is the program

There is no nacara binary. The engine is a library, your project references it, and Nacara.run gives that project a command line:

[<EntryPoint>]
let main argv = Nacara.run site argv

The engine version is a package reference, so the tool and the site cannot drift apart, and your configuration is ordinary F#: split it across files, share it between sites, generate it, or test it.

Where things are resolved from

Paths in the configuration - the collection source, the static directory, the output directory - are relative to the project root, which is the directory holding your site's project file.

All three of these do the same thing:

cd docs && dotnet run -- build
dotnet run --project docs/Docs.fsproj -- build
dotnet run --project /path/to/docs/Docs.fsproj -- build

--root <dir> overrides it, and --verbose prints the root it settled on. A published binary has no project file beside it, so it uses the current directory instead - pass --root to point it somewhere else.

Content and output

A collection reads files from one directory. Anything in the static directory is copied as it is. The output directory is rebuilt in place: files that nothing produces any more are deleted, so a page you delete does not stay online.