Seed Data and Collection Helpers

request.seed, map.*, and matrix.* — real runtime features for deterministic side-channel data, keyed bundles, and 2D calculations.

Seed Data and Collection Helpers#

The current runtime has practical, example-backed coverage for request.seed(...), map.*, and matrix.*. These are still partial parity surfaces compared with Pine, but they are no longer speculative or hidden implementation details.

request.seed#

Use it when the script needs deterministic side-channel data such as inline seeded series, structured local payloads, or reproducible auxiliary values that should remain in the frame-builder stage.

map and matrix#

Use map.* for keyed runtime bundles and matrix.* for compact 2D calculations. The docs site now treats both as supported teaching surfaces, not just footnotes in the capability matrix.

Example#

from source import indicator, input, map, matrix, request


indicator("Runtime Collections Demo", overlay=False)
window = input.int(4, title="Window", key="window", minval=2)


def build_indicator_frame(df, params=None):
    frame = df.copy().reset_index(drop=True)
    size = max(int((params or {}).get("window", window) or window), 2)

    seeded_source = {
        "seeded_close": frame["close"].rolling(size, min_periods=1).mean().round(4).tolist(),
    }
    frame["seeded_close"] = request.seed(seeded_source, "seeded_close")

    settings = map.new()
    map.put(settings, "window", size)
    map.put(settings, "bias", "bullish")
    frame.attrs["map_keys"] = map.keys(settings)

    grid = matrix.new(size, 2, 0.0)
    return frame

array / barstate / log

Libraries