Skip to content

Command line

cargo install dynars installs a dynars binary — a thin front end over the include-tree scanner. It's handy for a quick "what does this deck include, and how fast does it scan?" without writing any code, and its --json output drops straight into a shell pipeline.

cargo install dynars
dynars --help

dynars parse

Parse a root keyword file, walk its *INCLUDE graph, and print the tree plus scan throughput.

dynars parse root.k

Parsing: root.k
Threads: 10

=== Include Tree ===
root.k
  mesh/nodes.k
  mesh/shells.k
  materials.k
  ...

=== Performance ===
Total files:  128
Total bytes:  402653184 (384.00 MB)
Parse time:   0.026s (26.4ms)
Throughput:   14550.1 MB/s
              4848 files/s

Trees larger than 200 files print a summary line instead of the full listing.

JSON output

--json emits the whole tree plus timing as structured JSON — pipe it to jq, feed a dashboard, or diff two decks' file lists.

dynars parse root.k --json | jq '.total_files, .total_bytes'

{
  "root": "root.k",
  "total_files": 128,
  "total_bytes": 402653184,
  "parse_seconds": 0.0264,
  "throughput_mb_s": 14550.1,
  "tree": { "path": "root.k", "byte_count": 4096, "kind": null, "children": [ ... ] }
}

On a parse error, parse exits non-zero and writes the error (as {"error": …} under --json, or a plain Error: line otherwise) to stderr.

dynars generate

Generate a synthetic deck — a tree of *INCLUDEs full of *NODE lines — for benchmarking the scanner and the columnar path on decks of a chosen size and shape.

dynars generate --depth 6 --breadth 4 --nodes 100000 --output test_output
Flag Default Meaning
--depth 6 include nesting depth
--breadth 4 includes per file at each level
--nodes 100000 *NODE lines per file
--output test_output output directory

The resulting test_output/root.k is a normal deck — parse it with dynars parse, or point parse_deck at it from Python/Rust.

Beyond the CLI

The binary intentionally covers only the include-tree scan. For navigation, validation, columnar reads, and result files, use the library — start with Getting started. Validation makes an easy pre-submit gate; see gating a pipeline for a ready-to-adapt script.