fix: support 0D MeanView/WeightedMeanView

[Copilot]

Projecting all axes from a profile histogram yields a 0-d histogram, but accessing .values(), .variances(), .counts(), or any named field on the resulting view raised TypeError: iteration over a 0-d array.

h = bh.Histogram(bh.axis.Regular(10, 0, 100), storage=bh.storage.Mean())
h.fill(x, sample=y)

profile = h.project()          # 0-d Mean histogram
profile.values()               # TypeError: iteration over a 0-d array
profile.view()["value"]        # TypeError: iteration over a 0-d array
profile.view().value           # TypeError: iteration over a 0-d array

Root cause

In View.__getitem__, when a named field is accessed on a 0-d structured array (e.g. view["value"]), numpy returns a 0-d non-structured float array. The old code hit if not sliced.shape first and tried _PARENT._make(*sliced) — unpacking a 0-d array — before it could reach the dtype-changed guard.

Changes

• src/boost_histogram/view.py — swap check order in View.__getitem__: test sliced.dtype != self.dtype (field access) before not sliced.shape (scalar structured element). The shape check still correctly handles 0-d numpy void scalars produced by integer indexing on a 1-d view.
• tests/test_views.py — add test_0d_mean_view and test_0d_weighted_mean_view covering .values(), .variances(), .counts(), and direct field access on 0-d profiles.

---

📍 Connect Copilot coding agent with Jira, Azure Boards or Linear to delegate work to Copilot in one click without leaving your project management tool.