Top-level (NSE) support for weights and labels

[grantmcdermott]

Closes #332

Adds two new, top-level tinyplot() arguments:

• weights arg only passes through for the model-fit types (e.g., "lm", "glm") and distribution types (e.g., "histogram", "spineplot"). Passing through a top-level weights arg for a non-supported type is ignored, but triggers a warning (since it implies an unused statistical transformation).
• labels arg only passes through for type_text and is silently ignored otherwise.

Both of these are supported via NSE from the top-level, but users can also supply an equivalent atomic vector argument directly through the relevant type_*() constructor. For example, these two lines are equivalent:

plt(y ~ x, dat, type = "lm", weights = w)       # top-level, nse
plt(y ~ x, dat, type = type_lm(weights = dat$w) # type-level, vector

(However, the bottom variant is safer b/c it is incorporated into the model.frame computation, and so will correctly account for NAs across input columns, as well as subset operations.)

If both forms are provided at the same time, then the top-level variant wins out.

MWE

pkgload::load_all("~/Documents/Projects/tinyplot/")
#> ℹ Loading tinyplot
s77 = as.data.frame(state.x77)
plt(`Life Exp` ~ Income, data = s77)
plt_add(type = "lm")
plt_add(type = "lm", weights = Population, col = "red")

ttnc = as.data.frame(Titanic)
plt(
    Survived ~ Sex,
    data = ttnc,
    type = "spineplot",
    weights = Freq
)

aq = transform(airquality, mnth = month.abb[Month])
# plt(Ozone ~ Temp, data = aq, type = type_text(labels = aq$mnth)) ## would fail b/c of NAs in Ozone
plt(Ozone ~ Temp, data = aq, type = "text", labels = mnth)

^{Created on 2026-06-20 with reprex v2.1.1}

[grantmcdermott]

@zeileis @vincentarelbundock Not to nag or rush, but please let know if you have any conceptual objection to me merging this (or would like to test / kick the tyres before giving the go-ahead).

[zeileis]

I cannot keep up with your pace, Grant! 😳

But I gave this a quick trial now and noticed nothing problematic. In combination with other recent PRs we can now easily generate doubledecker plots from tabular input:

UCBAdmissions |>
  as.data.frame() |>
  tinyplot(Admit ~ Gender, facet = 1 ~ Dept, weights = Freq)

[grantmcdermott]

In combination with other recent PRs we can now easily generate doubledecker plots from tabular input:

Oh, that's a cool example. I'll add it to the documentation.