Enum EmbedderError 

pub enum EmbedderError {
    Download {
        model: String,
        source: String,
    },
    CacheUnwritable {
        source: String,
    },
    Load {
        model: String,
        source: String,
    },
    Inference {
        source: String,
    },
    EmbeddingsNotBuilt,
    DimensionMismatch {
        expected: usize,
        got: usize,
        model: String,
    },
    ModelMismatch {
        built: String,
        active: String,
    },
    Config {
        message: String,
    },
    NotCached {
        model: String,
        revision: Option<String>,
    },
}

A recoverable embedder failure. Returned instead of panicking so a load or inference problem surfaces to the SDK as a catchable error (with a remediation hint in Display) rather than aborting the host process.

Variants

Download

A configured embedding source could not be reached: offline, DNS/TLS, timeout, or an endpoint/model request returned a 4xx.

Fields

model: String

The configured model/source display name.

source: String

The underlying fetch error.

CacheUnwritable

The HuggingFace cache could not be written: permissions, disk full, or a read-only filesystem.

Fields

source: String

The underlying filesystem error.

Load

Model files are unusable: missing or corrupt weights, or a config/tokenizer that failed to parse.

Fields

model: String

The configured model/source display name.

source: String

The underlying load error.

Inference

Embedding a specific text failed (tokenization or the forward pass).

Fields

source: String

The underlying tokenizer/inference error.

EmbeddingsNotBuilt

A semantic/hybrid search was requested but the embedding cache is not built for the current corpus — build_embeddings was never run. No model is loaded; the caller must build the embeddings first.

DimensionMismatch

A vector’s dimension does not match the embedding cache’s. Cosine over mismatched dimensions is silently wrong, so this is a hard error — raised when a query (or an endpoint’s response) has a different width than the vectors the cache was built with.

Fields

expected: usize

Vector width the cache was built with.

got: usize

Vector width actually seen.

model: String

The active model, named for diagnosis.

ModelMismatch

A vector was produced by a different resolved model than the vectors already in the cache. Mixing vector spaces is never safe; callers must explicitly rebuild the full corpus to adopt the active model.

Fields

built: String

Resolved model identity that built the cache.

active: String

Resolved model identity returned by the active embedder.

Config

The embedding configuration is invalid — a bad source combination, a missing required field, or a named api_key_env that is not set. Surfaced at catalog construction or on first use.

Fields

message: String

What is wrong with the configuration.

NotCached

An explicitly-configured HuggingFace model is not in the local cache, and Ratel auto-downloads only the built-in default. The user must fetch it first — symmetric with Ollama’s “model not pulled”.

Fields

model: String

The repo id that is not cached.

revision: Option<String>

The requested revision, if pinned.

Trait Implementations

impl Clone for EmbedderError

fn clone(&self) -> EmbedderError

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for EmbedderError

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

Formats the value using the given formatter.

impl Display for EmbedderError

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

Formats the value using the given formatter.

impl Error for EmbedderError

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.