Defining Keys in a dataset

[koojo]

One Q which is going back to the basics would be does a key definition on a dataset enforce uniqueness?

I need to reuse a dataset multiple times. The keys defined while writhing the dataset are Key1, key2 & key3,

When the dataset is read the keys defined are Key1,Key2
and sometimes
key1, k2y2, key3, ke4 where key4 is another value field in the dataset.

In the read jobs I can use a copy stage to change the keys and read the data and then a join.

When I use the dataset to read data on Key1 and Key2 will I get a warning when used in a join?
i.e if I have duplicate data on these keys.

When I use key1,key2,key3 and key4 will there be any warnings when used in a join?

Is it best in this case for me to create three datasets for the three different keys and resue where I need them?