How does .impute interact with aggregated/transformed values?

[lucharo]

Hi, I recently came across an issue where in plotting time series values for missing dates where not being plotted but a line was was being drawn between them. In the absence of point=True the dots weren't being shown which made it difficult to perceive this at first. This resulted in the plot distorting the data, showing missing values that should be 0 as the interpolation of its neighbouring values.

I then researched altair docs for how to handle these cases and I came across
.impute which is great. My question is, how does .impute handle transformations in the encode stage? x=alt.X('sum(value)').impute(value = 0), does it impute the value column pre-aggregation or the transformation sum(value)?

find an interactive minimal reprex in this link: https://marimo.io/p/@lucharo/notebook-3cyj2l