matplotlib/matplotlib · error · TypeError

col must be an int or sequence of ints.

Error message

col must be an int or sequence of ints.

What it means

Table.auto_set_column_width wraps its argument with np.atleast_1d and requires the resulting dtype to be integer (table.py:508). A single float, a list of floats, string digits, or a float numpy array raises TypeError — even values like 1.0 that look like valid column indices.

Source

Thrown at lib/matplotlib/table.py:508

            ypos += heights[row]

        # set cell positions
        for (row, col), cell in self._cells.items():
            cell.set_x(lefts[col])
            cell.set_y(bottoms[row])

    def auto_set_column_width(self, col):
        """
        Automatically set the widths of given columns to optimal sizes.

        Parameters
        ----------
        col : int or sequence of ints
            The indices of the columns to auto-scale.
        """
        col1d = np.atleast_1d(col)
        if not np.issubdtype(col1d.dtype, np.integer):
            raise TypeError("col must be an int or sequence of ints.")
        for cell in col1d:
            self._autoColumns.append(cell)

        self.stale = True

    def _auto_set_column_width(self, col, renderer):
        """Automatically set width for column."""
        cells = [cell for key, cell in self._cells.items() if key[1] == col]
        max_width = max((cell.get_required_width(renderer) for cell in cells),
                        default=0)
        for cell in cells:
            cell.set_width(max_width)

    def auto_set_font_size(self, value=True):
        """Automatically set font size."""
        self._autoFontsize = value
        self.stale = True

View on GitHub (pinned to b379c1b69e)

Solutions

1. Pass ints: table.auto_set_column_width(0) or table.auto_set_column_width([0, 2])
2. Coerce untrusted input first: table.auto_set_column_width([int(c) for c in cols])
3. Generate candidate columns with np.arange(n), never np.linspace

Example fix

# before: float indices from a config file
auto_cols = [0.0, 2.0]
table.auto_set_column_width(auto_cols)  # TypeError: col must be an int or sequence of ints

# after: coerce to Python ints
table.auto_set_column_width([int(c) for c in auto_cols])

import numpy as np

def as_int_columns(col):
    a = np.atleast_1d(col)
    if not np.issubdtype(a.dtype, np.integer):
        if np.issubdtype(a.dtype, np.floating) and np.all(a == a.astype(int)):
            a = a.astype(int)
        else:
            raise TypeError(f'column indices must be ints, got dtype {a.dtype}')
    return a.tolist()

table.auto_set_column_width(as_int_columns(cfg['auto_cols']))

import numpy as np

def is_int_index_list(col) -> bool:
    try:
        a = np.atleast_1d(col)
    except Exception:
        return False
    return np.issubdtype(a.dtype, np.integer)

try:
    table.auto_set_column_width(col)
except TypeError as e:
    raise TypeError(f'auto_set_column_width needs int indices, got {col!r}') from e

When it happens

Trigger: table.auto_set_column_width(0.0); auto_set_column_width(np.array([0.0, 1.0])); auto_set_column_width(['0', '1']); column indices computed from divisions or read from JSON as floats.

Common situations: Indices derived from arithmetic (i / 2, np.linspace) or config/JSON parsing that yields floats; pandas .get_loc results assumed to be ints on categorical indexes (they can be slices/booleans).

Understand the failure class

Background: Invalid argument type errors: "must be of type string", "expected X, got Y", and ERR_INVALID_ARG_TYPE explained — this error's family across 15 libraries.

Related errors

• Image data of dtype {A.dtype} cannot be converted to float
• 'slope' cannot be used with non-linear scales
• the dtypes of parameters x ({x.dtype}) and width ({width.dty
• the dtypes of parameters y ({y.dtype}) and height ({height.d
• s must be float, but has type {s.dtype}

AI-assisted analysis of matplotlib/matplotlib@b379c1b69e (2026-08-21). Data as JSON: /api/errors/36c4bcc7dd4e3cdc. Report an issue: GitHub.