Amalgamation in condensed scores

I want to condense soprano/alto in a choir.
Sometimes the different voices splits up so I’ll have two alto (a1 & a2) and one soprano voice.
With this in mind I prefer all soprano notes with stem up and all alto notes with stem down I think it’s much easier to read.
To do this I select Notation Options: Prevent amalgamation

but my sopranos and alto voices is still both written with stems up.
Look here:

Bar 1 and 2 are ok, but bar 3 and 4 is not.

Is it a bug? Or do I have to do something else

Read exactly what that first screenshot says at the bottom (it’s perfectly clear that your result isn’t a bug), and then resort to Manual Condensing (if necessary).

Amalgamation is to use single stems where possible if the phrase has different rhythms among voices. After the rest, a new phrase starts, but it’s homorhythmic, so amalgamation isn’t actually happening: it’s just single stem condensing.

To get what you want, you can put a condensation change at the beginning of the phrase (or the whole piece), enable Manual Condensing, and put Soprano in the upstem voice and Alto in the downstem voice.

In future we plan to add an option specifically for vocal staves that would prevent Dorico from getting as far as a single shared stem for the voices sharing a staff, since that is a convention that is used, of course, but it’s just as easy to use manual condensing to produce this result.

Okay now I understand!
Thanks a lot.