Struct TensorDesc 

pub struct TensorDesc {
    pub name: Option<String>,
    pub optional: bool,
    pub dtypes: Vec<String>,
    pub dtype_class: Option<String>,
    pub layout: Option<LayoutSpec>,
    pub rank: Option<Value>,
    pub shape_constraint: Option<String>,
    pub fdx: Option<FdxSpec>,
    pub device: Option<String>,
    pub substrate: Option<String>,
}

A tensor descriptor (operand or, with return rules, an output; §3.2).

Fields

name: Option<String>

Operand role name (diagnostic + maps to an FDX view name, §5.5).

optional: bool

true ⇒ this INPUT operand is OPTIONAL (§3.4 optional-operand support): production registers the op BOTH with and without it (e.g. conv’s bias). The importer’s key-builder fans each dtype variant into two keys — one omitting this operand, one including it — when set. Only the LAST input may be optional. Defaults to false (a required operand); #[serde(default)] so every existing contract that omits the field keeps building exactly one key per dtype variant (additive §11).

dtypes: Vec<String>

Accepted DLPack dtype names (Fuel DType names; FDX §6.1). Strings; the string → DType conversion is a lowering concern.

dtype_class: Option<String>

Optional shorthand int|uint|float|any that expands at lower time (§3.4).

layout: Option<LayoutSpec>

Layout capability flags (§4.1).

rank: Option<Value>

Rank: exact int, "any", or a range "2..=4". Kept as a raw scalar so 2 and "any" both parse (the FKC parser interprets it later).

shape_constraint: Option<String>

Free-predicate shape constraint (§3.5). Carried verbatim.

fdx: Option<FdxSpec>

DLPack-extension (FDX) requirements (§3.2 fdx:).

device: Option<String>

Placement device (inherited from front-matter unless overridden).

substrate: Option<String>

FDX substrate class.

Trait Implementations

impl Clone for TensorDesc

fn clone(&self) -> Self

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Debug for TensorDesc

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl<'de> Deserialize<'de> for TensorDesc

fn deserialize<__D>(__deserializer: __D) -> Result<Self, __D::Error>

where __D: Deserializer<'de>,

Deserialize this value from the given Serde deserializer.