Actually, the following method always returns a positive integer:
var fixture = new Fixture();
var someInt = fixture.Create<int>();
Is it possible that one day, the feature evolves and begins to return a negative integer? Or is there an official reason it actually returns an always positive integer
If not, is there some documentation hidden somewhere telling it?
To keep on using autofixture, I'd like to know if some changes are foreseen.
The author of Autofixture discusses this on his blog. This post specifies that the current implementation will always return positive numbers since they are deemed "safer" in general, so I don't think this will change in the near future.
The whole point of AutoFixture is to generate anonymous test data. You are asking for an integer which can be a negative number. To be 100% safe, I wouldn't rely on the implicit assumption that all future implementations only return positive numbers. You can make this more explicit by providing a custom SpecimenBuilder:
fixture.Customizations.Add(new PositiveIntegerBuilder());
More information about custom specimen builders can be found here.
As user prgmtc points out, one option is through a custom ISpecimenBuilder.
Another option is to supply a custom range, by using the built-in RandomNumericSequenceGenerator class, as shown below:
[Fact]
public void FixtureCreatesNegativeNumbers()
{
var fixture = new Fixture();
fixture.Customizations.Add(
new RandomNumericSequenceGenerator(-900, -100));
var i = fixture.Create<int>();
// Prints -> -711
var l = fixture.Create<long>();
// Prints -> 618
var f = fixture.Create<float>();
// Prints -> -78.0
}
var i = fixture.Create<uint>();
Only if the original code being tested expected a uint type parameter.
Wouldvar i = Math.Abs(fixture.Create<int>()) work for you
To answer the original question
I would assume that the authors of AutoFixture will not change something that crucial so lightly.
I would use fixture.Create<int>