Graphrule Aggfunc multiple aggregations

[Zap123]

I am in need to aggregate my stream on multiple columns, in SQL I would do something similar to
SELECT AVG(temperature), MIN(temperature).
I've tried having a node with an expression equal to the SQL version like so "AVG(temperature), MIN(temperature)" but it seems that only the average get returned.
After many tries I've tried splitting each aggregation in a different aggfunc node and this way it works but it doesn't make sense to me
The documentation says:
"The input for the node must be a collection of rows as the output of a window. The node will aggregate multiple rows into one aggregated row. "
So semantically I shouldn't be able to link two aggregations because I'm expecting a row. After digging through the source code I've seen that there's a isAgg variable so I guess the engine combine them somehow.

Is this to be expected or did I made a mistake in my flow that prevent it to work with one single node? The previous node is a window node as I would like to compute avg and min from a 10 minutes stream.
I haven't found it mentioned anywhere in the documentation.

[ngjaying]

@Zap123 Initially, the Graph API was based on SQL concepts. However, we intentionally moved away from this design in later versions because it was counterintuitive for users. To address this, we made each operator compatible with both scalar values and collections so that users can actually connect any nodes.

This change means that when an aggregation function runs, the new aggregated value is simply appended to the existing data. Because the original collection information is preserved, you can perform another aggregation in a subsequent node.

[Zap123]

Thanks @ngjaying this clarifies it. I like the Graph API a lot but unfortunately I couldn't find this detail in the documentation and I had to discover it.