Alteration to unstack a result-set

[Seddryck]

It should be possible unstack (or pivot) a result-set. A first set of columns (key-columns) is used to determine which row should be grouped together. Each unique group of values for the key-columns will create a new row in the resulting result-set. The distinct values of a column (the header-column) become columns of the resulting dataset.

<alteration>
  <unstack>
    <header>
      <column identifier="myHeaderColumn"/>
    </header>
    <group-by>
      <column identifier="myKeyColumn" type="text"/>
    </group-by>
  </unstack>
</alteration>

It means that the following table

Key1	Header1	Value1
one	a	1.0
one	b	2.0
two	a	3.0
two	b	4.0

will be reshapped to

Key1	a	b
one	1.0	2.0
two	3.0	4.0

If more than one value column is present then the resulting header will be the concatenation of the header column and the value column name.

Key1	Header1	Value1	Value2
one	a	1.0	-1.0
one	b	2.0	-2.0
two	a	3.0	-3.0
two	b	4.0	-4.0

will be reshaped to

Key1	a_Value1	b_Value1	a_Value2	b_Value2
one	1.0	2.0	-1.0	-2.0
two	3.0	4.0	-3.0	-4.0

If at any moment, two rows of the initial table try to fill the same cell of the resulting table, then the reshapping is failing.

[Seddryck]

Beta available in 1.21.0-beta0313