DuckDB-WASM
    Preparing search index...

    Interface DuckDBConfig

    interface DuckDBConfig {
        accessMode?: DuckDBAccessMode;
        allowUnsignedExtensions?: boolean;
        arrowLosslessConversion?: boolean;
        customUserAgent?: string;
        filesystem?: DuckDBFilesystemConfig;
        maximumThreads?: number;
        opfs?: DuckDBOPFSConfig;
        path?: string;
        query?: DuckDBQueryConfig;
        singleThreadedExtensionLoad?: boolean;
        useDirectIO?: boolean;
    }
    Index

    Properties

    accessMode?: DuckDBAccessMode

    The access mode

    allowUnsignedExtensions?: boolean

    Whether to allow unsigned extensions

    arrowLosslessConversion?: boolean

    Whether to use alternate Arrow conversion that preserves full range and precision of data.

    customUserAgent?: string

    Custom user agent string

    The filesystem config

    maximumThreads?: number

    The maximum number of threads. Note that this will only work with cross-origin isolated sites since it requires SharedArrayBuffers.

    opfs string

    path?: string

    The database path

    The query config

    singleThreadedExtensionLoad?: boolean

    Run extension LOAD/INSTALL statements with the task scheduler pinned to a single thread.

    On a cross-origin-isolated (threaded) build the engine fetches an extension binary with a blocking synchronous XHR. With background scheduler threads live, that blocking fetch can deadlock across the worker/main-thread boundary (and exhaust the emscripten pthread pool). When enabled, the binding brackets each LOAD/INSTALL with SET threads=1SET threads=<previous> so the fetch runs with no background workers, then restores the prior thread count.

    Defaults to enabled. It is a no-op on single-threaded (mvp/eh) builds and whenever threads is already 1, so it only ever engages on threaded builds that actually have background workers. Set to false to opt out.

    useDirectIO?: boolean

    The direct io flag