`value` of `selection_interval` should be list, not tuple

[rambip]

What happened?

I wanted to create a visualization of a 2D grid, with an initial selection defined as a subgrid of the larger grid.
Here is a minimal reproducible example (y axis only):

import polars as pl
import altair as alt
# Create data
data = pl.DataFrame([
    {"x": "1", "y": "A", "value": 1},
    {"x": "2", "y": "A", "value": 2},
    {"x": "3", "y": "A", "value": 3},
    {"x": "1", "y": "B", "value": 4},
    {"x": "2", "y": "B", "value": 5},
    {"x": "3", "y": "B", "value": 6},
    {"x": "1", "y": "C", "value": 7},
    {"x": "2", "y": "C", "value": 8},
    {"x": "3", "y": "C", "value": 9},
    {"x": "1", "y": "D", "value": 10},
    {"x": "2", "y": "D", "value": 11},
    {"x": "3", "y": "D", "value": 12},
])

# Create interval selection with initial value
brush = alt.selection_interval(
    encodings=['y'],
    value={'y': ['A', 'B', 'C']}
)

# Create heatmap
chart = alt.Chart(data).mark_rect().encode(
    x=alt.X('x:N'),
    y=alt.Y('y:N'),
    color=alt.Color('value:Q', scale=alt.Scale(scheme='blues')),
    opacity=alt.condition(brush, alt.value(1), alt.value(0.3))
).add_params(
    brush
).properties(
    width=400,
    height=300,
)

chart

But it does not work. I get

Multiple errors were found. Error 1: 'A' is an invalid value for `0`. Valid values are of type 'boolean', 'number', or 'object'. Error 2: 'B' is an invalid value for `1`. Valid values are of type 'boolean', 'number', or 'object'. Error 3: 'C' is an invalid value for `2`. Valid values are of type 'boolean', 'number', or 'object'.

At this point: I am confused for 2 reasons:

• the error message is not helpful at all to know why it fails (I tried to do it again with number instead of letters, and same error).
• Should the content of the list be the interval (start-end value) or the list of all selected values ? The documentation does not give any hint (https://altair-viz.github.io/user_guide/generated/api/altair.selection_interval.html)

What would you like to happen instead?

I was able to implement what I wanted in Vega-lite, and it actually works:

{
      "$schema": "https://vega.github.io/schema/vega-lite/v5.json",
      "width": 400,
      "height": 300,
      "data": {
        "values": [
          {"x": "1", "y": "1", "value": 1},
          {"x": "2", "y": "1", "value": 2},
          {"x": "3", "y": "1", "value": 3},
          {"x": "1", "y": "2", "value": 4},
          {"x": "2", "y": "2", "value": 5},
          {"x": "3", "y": "2", "value": 6},
          {"x": "1", "y": "3", "value": 7},
          {"x": "2", "y": "3", "value": 8},
          {"x": "3", "y": "3", "value": 9},
          {"x": "1", "y": "4", "value": 10},
          {"x": "2", "y": "4", "value": 11},
          {"x": "3", "y": "4", "value": 12},
        ]
      },
      "params": [{
        "name": "brush",
        "select": {
          "type": "interval",
          "encodings": ["y"]
        },
        "value": {"y": ["1", "2", "3"]}
      }],
      "mark": "rect",
      "encoding": {
        "x": {"field": "x", "type": "ordinal"},
        "y": {"field": "y", "type": "ordinal"},
        "color": {
          "field": "value",
          "type": "quantitative",
          "scale": {"scheme": "blues"}
        },
        "opacity": {
          "condition": {"param": "brush", "value": 1},
          "value": 0.3
        }
      }
}

Output:

Altair should show exactly the same thing, without the weird error.

Note: I don't know if the specification somewhere, but there might be a mismatch between the vega-lite spec and the current way user understand the parameter (see this post as an example)

Which version of Altair are you using?

5.5.0

[mattijn]

I can reproduce this error using this:

import altair as alt

alt.selection_interval(
    encodings=['y'],
    value={'y': ['A', 'B', 'C']}
)

It is something we should be able to support, and is related to types in a part of the repository which is not auto generated.

Assistance with this issue regarding types is appreciated!