Enum LlmCallError 

pub enum LlmCallError {
    NotConfigured {
        model: String,
    },
    UnknownModel {
        model: String,
        available: Vec<String>,
    },
    BudgetExceeded {
        model: String,
        limit_kind: LlmLimitKind,
    },
    PromptBoundsExceeded {
        model: String,
        limit_kind: PromptBoundKind,
        actual: u64,
        limit: u64,
    },
    Unauthorized {
        model: String,
    },
    Transport {
        model: String,
        detail: String,
    },
}

Why a model dispatch failed — the request never reached the model, or the transport died before any element ran.

Reserved for dispatch-level failure only. A model that answered and produced an unusable or incomplete element reports that in LlmOutcome’s failure arm, never here: returning Err for one bad element would throw away the sibling elements that did succeed.

Every variant names the model it was dispatched at and the fix the caller can act on. None carries prompt or completion text.

Variants

NotConfigured

No LlmProvider is wired into the runtime, so nothing can dispatch. The pure-interpreter path leaves the provider None, and this is the catchable configuration error a script sees there.

Fields

model: String

UnknownModel

The model name is not one the provider serves. available lists what it does serve, so the message can name the alternatives rather than only the rejection.

Fields

model: String

available: Vec<String>

BudgetExceeded

A token ceiling cannot cover this dispatch. Charged before any element ran, so nothing was billed.

limit_kind carries which ceiling refused, because the three want different responses and a message that reads the same for all of them sends a caller to shrink work that was never what filled the budget.

Fields

model: String

limit_kind: LlmLimitKind

PromptBoundsExceeded

The prompt slice is outside the bounds the host accepts — too many elements, or one element too large. Counted in elements and bytes, never quoted.

Fields

model: String

limit_kind: PromptBoundKind

actual: u64

limit: u64

Unauthorized

The provider rejected the credential, or none was resolvable when the provider was constructed. One variant covers both because the fix is the same key — this is the LoadAPIKeyError-vs-401 split collapsed at the point where the caller can act on it.

Fields

model: String

Transport

The request never completed, and died before any element ran. A transport failure after elements started belongs in FailureReason::Transport on the affected elements instead.

detail is a fixed classification string — “connection reset”, “dns lookup failed”, “tls handshake failed”. It is the one free-form field in this taxonomy, and the no-payload rule binds it: never a provider response body, and never a request echo. The precedent this trait otherwise mirrors does interpolate a body (McpCallError::Upstream renders HTTP {status}: {body}); this one deliberately does not, because a model provider’s body carries completion fragments.

Fields

model: String

detail: String

Implementations

impl LlmCallError

pub fn model(&self) -> &str

The model this dispatch was aimed at. Every variant names one, so a caller need not match to attribute the failure.

pub fn is_budget_exceeded(&self) -> bool

Token-budget refusals become catchable QuotaExceededErrors at the stdlib boundary. Prompt size/count bounds remain RangeErrors.

Trait Implementations

impl Clone for LlmCallError

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 LlmCallError

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

Formats the value using the given formatter.

impl Display for LlmCallError

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

Formats the value using the given formatter.

impl Eq for LlmCallError

impl Error for LlmCallError

fn source(&self) -> Option<&(dyn Error + 'static)>

Returns the lower-level source of this error, if any.

fn description(&self) -> &str

👎Deprecated since 1.42.0:

use the Display impl or to_string()

fn cause(&self) -> Option<&dyn Error>

👎Deprecated since 1.33.0:

replaced by Error::source, which can support downcasting

fn provide<'a>(&'a self, request: &mut Request<'a>)

🔬This is a nightly-only experimental API. (error_generic_member_access)

Provides type-based access to context intended for error reports.

impl PartialEq for LlmCallError

fn eq(&self, other: &Self) -> bool

Equality operator ==.

fn ne(&self, other: &Rhs) -> bool

Inequality operator !=.

impl StructuralPartialEq for LlmCallError