langchain-ai/langchain · warning · ValueError

Canvas dimensions should be > 1

Error message

Canvas dimensions should be > 1

What it means

`AsciiCanvas.__init__` (used by `DrawGraph` to render runnable graphs as ASCII art) requires both `cols` and `lines` to be greater than 1, otherwise there is no drawable area. Violating this raises `ValueError: Canvas dimensions should be > 1`.

Source

Thrown at libs/core/langchain_core/runnables/graph_ascii.py:74

class AsciiCanvas:
    """Class for drawing in ASCII."""

    TIMEOUT = 10

    def __init__(self, cols: int, lines: int) -> None:
        """Create an ASCII canvas.

        Args:
            cols: number of columns in the canvas. Should be `> 1`.
            lines: number of lines in the canvas. Should be `> 1`.

        Raises:
            ValueError: if canvas dimensions are invalid.
        """
        if cols <= 1 or lines <= 1:
            msg = "Canvas dimensions should be > 1"
            raise ValueError(msg)

        self.cols = cols
        self.lines = lines

        self.canvas = [[" "] * cols for line in range(lines)]

    def draw(self) -> str:
        """Draws ASCII canvas on the screen.

        Returns:
            The ASCII canvas string.
        """
        lines = map("".join, self.canvas)
        return os.linesep.join(lines)

    def point(self, x: int, y: int, char: str) -> None:
        """Create a point on ASCII canvas.

View on GitHub (pinned to e32fa9a52e)

Solutions

1. Clamp dimensions to a minimum: `max(cols, 2)` and `max(lines, 2)` before constructing.
2. Use the default `DrawGraph(graph).draw()` path instead of constructing `AsciiCanvas` with computed sizes.
3. If customizing `DrawGraph`, keep `node_size` and the derived canvas math well above 1.

Example fix

# before
canvas = AsciiCanvas(int(width * ratio), int(height * ratio))  # can hit 1

# after
canvas = AsciiCanvas(max(int(width * ratio), 2), max(int(height * ratio), 2))

cols, lines = max(cols, 2), max(lines, 2)
canvas = AsciiCanvas(cols, lines)

When it happens

Trigger: Instantiating `AsciiCanvas(0, 10)`, `AsciiCanvas(80, 1)`, or with negative values; calling `AsciiCanvas`-based rendering with dimensions computed from graph size (e.g. `node_degree * scale`) where the computation yields 0/1 for tiny graphs.

Common situations: Auto-sizing logic that scales canvas size by the number of nodes/edges and collapses to <=1 for a minimal graph (single node, no edges); passing terminal width from a CI environment where `shutil.get_terminal_size()` reports tiny dimensions; customizing `DrawGraph` node size constants to 0.

Related errors

• char should be a single character
• x should be >= 0 and < number of columns
• Node with id {id} already exists
• Source node {source.id} not in graph
• Target node {target.id} not in graph

AI-assisted analysis of langchain-ai/langchain@e32fa9a52e (2026-08-14). Data as JSON: /api/errors/ddfd8469aafb454d. Report an issue: GitHub.