.facet does not accept facet=alt.Facet('x', columns=2) as an input

[samesand]

.facet accepts alt.Facet('x') as a input, but if you add columns as a parameter to alt.Facet it returns a 'data' is a required property error. I think this happens because .facet accepts columns as an argument as well.

Example:
alt.Chart(df).mark_circle().encode(x='x', y='y').facet(facet=alt.Facet('z', columns=2)

If this is not a bug, maybe the documentation for alt.Facet should indicate the expected use of the columns argument (or other arguments that may not work when used in the same manner)

[joelostblom]

Thanks for reporting this issue! Related SO question here for reference with some discussion in the comments

[joelostblom]

This is now a bit more confusing because the new improved error message suggests that columns should work with Facet:

import altair as alt
from vega_datasets import data

source = data.barley.url

alt.Chart(source).mark_point().encode(
    alt.X('median(yield):Q').scale(zero=False),
    y='variety:O',
    color='year:N'
).properties(
    width=200,
    height=100,
).facet(
    alt.Facet('site:O', columns=2),
)

Error message contradicting itself since columns is listed in the existing parameter names:

SchemaValidationError: `Facet` has no parameter named 'columns'

Existing parameter names are:
shorthand      bin       field     timeUnit   
aggregate      bounds    header    title      
align          center    sort      type       
bandPosition   columns   spacing              

See the help for `Facet` to read the full description of these parameters

[dangotbanned]

This isn't a bug, the issue was how many (f|F)acet variations were used

import polars as pl
import altair as alt

df = pl.DataFrame({"x": [1, 2, 3], "y": [1, 2, 3], "z": [1, 2, 3]})
chart = alt.Chart(df).mark_circle().encode(x="x", y="y")

Error

chart.facet(facet=alt.Facet('z', columns=2))

Traceback

SchemaValidationError: `Facet` has no parameter named 'columns'

Existing parameter names are:
shorthand      bin       field     timeUnit   
aggregate      bounds    header    title      
align          center    sort      type       
bandPosition   columns   spacing              

See the help for `Facet` to read the full description of these parameters

Solution

chart.facet("z", columns=2)

Screenshot

[joelostblom]

The issue is that when you do something like chart.facet(alt.Facet('z', columns=2)), the error message says

SchemaValidationError: `Facet` has no parameter named 'columns'

But then proceeds to list columns as a valid parameter name for alt.Facet:

Existing parameter names are:
shorthand      bin         field     timeUnit   
aggregate      bounds      header    title      
align          center      sort      type       
bandPosition   **columns** spacing              

See the help for `Facet` to read the full description of these parameters

It's not a huge deal, but somewhat confusing and does not point users in the right direction (columns outside alt.Facet as in your solution).

[dangotbanned]

It's not a huge deal, but somewhat confusing and does not point users in the right direction (columns outside alt.Facet as in your solution).

@joelostblom I do see the error message as being an issue, maybe something to look into as part of #3752?

Its tricky because Facet in this example could refer to either:

import altair as alt
alt.Chart().facet(alt.Facet('z', columns=2))
#           ^^^^^

Or:

import altair as alt
alt.Chart().facet(alt.Facet('z', columns=2))
#                     ^^^^^

[joelostblom]

Maybe something to look into as part of #3752?

That sounds like a good idea!