From Spock documentation:
Data tables, data pipes, and variable assignments can be combined as needed:
... where: a | _ 3 | _ 7 | _ 0 | _ b << [5, 0, 0] c = a > b ? a : b
This simple example produces a MissingPropertyException
for y
.
def test() {
expect:
x == 42
where:
y = 12
x << [y + 30, 54 - y]
}
What's wrong with this example?
You can't mix data pipe and variable assignment that way. Let's take a look at the source code. You can put a breakpoint in org.spockframework.runtime.ParameterizedSpecRunner
class at line 49 and run the test with the debugger. You will see that currentFeature
holds two parameters with names y
and x
:
You will also notice that there is a single dataProvider
for variable name x
defined:
This data provider exists because we have defined x
as a data pipe, so it has to iterate over the list of variables and evaluate it in the context of data pipes. In this context it expects that y
variable is also defined as a data pipe so it can take a value associated with the same index.
If you define your where:
as:
where:
y << [12, 12]
x << [y + 30, 54 - y]
your test would succeed, because now y
variable exists as a data provider and evaluating values for x
accesses values for y
using second data provider.
Consider following example:
@Unroll
def "variable assignment example"() {
expect:
x == 42
and:
c
where:
y << [12, 12]
x << [y + 30, 54 - y]
c = y < x
}
In this case variable c
is evaluated twice for each element in data pipes y
and x
. When the first unroll happens y = 12
, x = y + 30 => 12 + 30 => 42
and c
evaluates to true
because y<x
. When the second unroll happens 12
value is assigned to y
again, x
evaluates to 42
(54 - y => 54 - 12 => 42
) and c
evaluates to true
again.
I see that it may look like simple y = 12
variable assignment should be discovered by data pipe evaluation. Unfortunately it does not work that way. When data pipe gets evaluated it can only use values from other data pipes and any variable assignment is not seen in this context.