[MNT]: Consistency on Interface

[suuuehgi]

Summary

D = {ii:ii for ii in range(10)}

fig, ax = plt.subplots()
plt.errorbar(D.keys(), D.values(), D.values())

raises the error

TypeError: unsupported operand type(s) for *: 'int' and 'dict_values'

while

fig, ax = plt.subplots()
ax.errorbar(D.keys(), D.values(), D.values())

works as expected.

(mpl v. 3.5.3)

Proposed fix

Both work.

[tacaswell]

I get a slightly different error with mpl3.7.

File /usr/lib/python3.10/site-packages/matplotlib/pyplot.py:2564, in errorbar(x, y, yerr, xerr, fmt, ecolor, elinewidth, capsize, barsabove, lolims, uplims, xlolims, xuplims, errorevery, capthick, data, **kwargs)
   2558 @_copy_docstring_and_deprecators(Axes.errorbar)
   2559 def errorbar(
   2560         x, y, yerr=None, xerr=None, fmt='', ecolor=None,
   2561         elinewidth=None, capsize=None, barsabove=False, lolims=False,
   2562         uplims=False, xlolims=False, xuplims=False, errorevery=1,
   2563         capthick=None, *, data=None, **kwargs):
-> 2564     return gca().errorbar(
   2565         x, y, yerr=yerr, xerr=xerr, fmt=fmt, ecolor=ecolor,
   2566         elinewidth=elinewidth, capsize=capsize, barsabove=barsabove,
   2567         lolims=lolims, uplims=uplims, xlolims=xlolims,
   2568         xuplims=xuplims, errorevery=errorevery, capthick=capthick,
   2569         **({"data": data} if data is not None else {}), **kwargs)

File /usr/lib/python3.10/site-packages/matplotlib/__init__.py:1442, in _preprocess_data.<locals>.inner(ax, data, *args, **kwargs)
   1439 @functools.wraps(func)
   1440 def inner(ax, *args, data=None, **kwargs):
   1441     if data is None:
-> 1442         return func(ax, *map(sanitize_sequence, args), **kwargs)
   1444     bound = new_sig.bind(ax, *args, **kwargs)
   1445     auto_label = (bound.arguments.get(label_namer)
   1446                   or bound.kwargs.get(label_namer))

File /usr/lib/python3.10/site-packages/matplotlib/axes/_axes.py:3640, in Axes.errorbar(self, x, y, yerr, xerr, fmt, ecolor, elinewidth, capsize, barsabove, lolims, uplims, xlolims, xuplims, errorevery, capthick, **kwargs)
   3635     raise ValueError(
   3636         f"'{dep_axis}err' (shape: {np.shape(err)}) must be a "
   3637         f"scalar or a 1D or (2, n) array-like whose shape matches "
   3638         f"'{dep_axis}' (shape: {np.shape(dep)})") from None
   3639 res = np.zeros(err.shape, dtype=bool)  # Default in case of nan
-> 3640 if np.any(np.less(err, -err, out=res, where=(err == err))):
   3641     # like err<0, but also works for timedelta and nan.
   3642     raise ValueError(
   3643         f"'{dep_axis}err' must not contain negative values")
   3644 # This is like
   3645 #     elow, ehigh = np.broadcast_to(...)
   3646 #     return dep - elow * ~lolims, dep + ehigh * ~uplims
   3647 # except that broadcast_to would strip units.

TypeError: bad operand type for unary -: 'dict_values'

@suuuehgi In the future please post the full traceback, it makes it much easier to understand what the problem may be.

Given that the body of plt.errorbar just forwards everything to ax.errorbar this is a bit surprising, however I can also produce this error via :

ax.errorbar(D.keys(), D.values(), xerr=D.values())

so we are apparently doing some normalization on *args that we are not doing on keywords?

Labeling as good first issue because there is no API design, but medium difficulty as it will involve understanding the input validation in errorbar which I do not think is trivial.

[suuuehgi]

@suuuehgi In the future please post the full traceback, it makes it much easier to understand what the problem may be.

Of course, sorry!

---------------------------------------------------------------------------
TypeError                                 Traceback (most recent call last)
Cell In[21], line 2
      1 fig, ax = plt.subplots()
----> 2 plt.errorbar(D.keys(), D.values(), D.values())

File ~/.local/lib/python3.10/site-packages/matplotlib/pyplot.py:2519, in errorbar(x, y, yerr, xerr, fmt, ecolor, elinewidth, capsize, barsabove, lolims, uplims, xlolims, xuplims, errorevery, capthick, data, **kwargs)
   2513 @_copy_docstring_and_deprecators(Axes.errorbar)
   2514 def errorbar(
   2515         x, y, yerr=None, xerr=None, fmt='', ecolor=None,
   2516         elinewidth=None, capsize=None, barsabove=False, lolims=False,
   2517         uplims=False, xlolims=False, xuplims=False, errorevery=1,
   2518         capthick=None, *, data=None, **kwargs):
-> 2519     return gca().errorbar(
   2520         x, y, yerr=yerr, xerr=xerr, fmt=fmt, ecolor=ecolor,
   2521         elinewidth=elinewidth, capsize=capsize, barsabove=barsabove,
   2522         lolims=lolims, uplims=uplims, xlolims=xlolims,
   2523         xuplims=xuplims, errorevery=errorevery, capthick=capthick,
   2524         **({"data": data} if data is not None else {}), **kwargs)

File ~/.local/lib/python3.10/site-packages/matplotlib/__init__.py:1414, in _preprocess_data.<locals>.inner(ax, data, *args, **kwargs)
   1411 @functools.wraps(func)
   1412 def inner(ax, *args, data=None, **kwargs):
   1413     if data is None:
-> 1414         return func(ax, *map(sanitize_sequence, args), **kwargs)
   1416     bound = new_sig.bind(ax, *args, **kwargs)
   1417     auto_label = (bound.arguments.get(label_namer)
   1418                   or bound.kwargs.get(label_namer))

File ~/.local/lib/python3.10/site-packages/matplotlib/axes/_axes.py:3464, in Axes.errorbar(self, x, y, yerr, xerr, fmt, ecolor, elinewidth, capsize, barsabove, lolims, uplims, xlolims, xuplims, errorevery, capthick, **kwargs)
   3456     raise ValueError(
   3457         f"'{dep_axis}err' (shape: {np.shape(err)}) must be a "
   3458         f"scalar or a 1D or (2, n) array-like whose shape matches "
   3459         f"'{dep_axis}' (shape: {np.shape(dep)})") from None
   3460 # This is like
   3461 #     elow, ehigh = np.broadcast_to(...)
   3462 #     return dep - elow * ~lolims, dep + ehigh * ~uplims
   3463 # except that broadcast_to would strip units.
-> 3464 low, high = dep + np.row_stack([-(1 - lolims), 1 - uplims]) * err
   3466 barcols.append(lines_func(
   3467     *apply_mask([indep, low, high], everymask), **eb_lines_style))
   3468 # Normal errorbars for points without upper/lower limits.

TypeError: unsupported operand type(s) for *: 'int' and 'dict_values'

I initially had this issue with the x argument but for reasons I could not reproduce it.

[ksunden]

matplotlib/lib/matplotlib/cbook.py

Line 1663 in a844eca

def sanitize_sequence(data):

The sanitize_sequence method (which is called by the _preprocess_data decorator) explicitly casts MappingView to list, but only for args (not kwargs)

The difference then comes down to how a mapping view vs list are handled by np.asarray:

>>> np.asarray({"a": 1, "b": 2}.values(), dtype=object)
array(dict_values([1, 2]), dtype=object)
>>> np.asarray({"a": 1, "b": 2}.values(), dtype=object).shape
()
>>> np.asarray([1,2], dtype=object)
array([1, 2], dtype=object)
>>> np.asarray([1,2], dtype=object).shape
(2,)

So the dict view is a scalar of the view object, the list is a 1-d array.

The line that is calling asarray, with dtype of object:

matplotlib/lib/matplotlib/axes/_axes.py

Line 3506 in a844eca

return np.asarray(err, dtype=object)

[anntzer]

FWIW I think special-casing support for MappingViews was a mistake (if we really want to support them we should lobby numpy to support np.asarray(mappingview) with the semantics we want) and would support just deprecating that (we can easily emit a warning saying "please convert to a list with list(view)").