Errors

Errors

The public Result<T> alias wraps the crate’s Error enum. The variants come from src/core/error.rs.

Error variants#

Variant	When it appears
TokenBudgetExceeded	Public error variant for budget failures; the current builder path usually selects or truncates instead of raising it.
InvalidConfig	A provided configuration is not valid for the current operation.
ChunkingError	Chunk generation or parsing failed.
SerializationError	Serde failed while parsing or writing JSON.
TokenCountError	Tokenizer initialization or counting failed.
MemoryError	Reserved public variant for memory-related failures.
ContextBuildError	General context assembly or model setup problem.
Embedding	Embedding generation failed.
Model	Local text model initialization or inference failed.

Handling errors#

rust

use forgetless::{Error, Forgetless};

match Forgetless::new().add("content").run().await {
    Ok(result) => println!("{}", result.total_tokens),
    Err(Error::TokenBudgetExceeded { requested, available }) => {
        eprintln!("budget exceeded: {requested} > {available}");
    }
    Err(Error::InvalidConfig(message)) => eprintln!("invalid config: {message}"),
    Err(Error::ChunkingError(message)) => eprintln!("chunking failed: {message}"),
    Err(Error::TokenCountError(message)) => eprintln!("tokenizer failed: {message}"),
    Err(Error::ContextBuildError(message)) => eprintln!("context build failed: {message}"),
    Err(Error::Embedding(message)) => eprintln!("embedding failed: {message}"),
    Err(Error::Model(message)) => eprintln!("model failed: {message}"),
    Err(other) => eprintln!("{other}"),
}

Configuration

Configure Forgetless token budgets, scoring weights, chunk behavior, local model options, and feature toggles to tune LLM context for your workload.

Tokenizer

Count LLM tokens in Rust with cl100k_base or a custom chars-per-token mode, plus truncation helpers and image token estimation to plan context budgets.

Results

Read the Forgetless result types: OptimizedContext, selected chunks, score breakdowns, token statistics, and the compression ratio returned after each run.