More Testing Tips With C#

The post gathers a few C# testing tips around test data, reference outputs, and integration-test setup. It says test data generators help when hand-written data gets tedious, seeds make randomized tests reproducible, and reference files work well for stable rendering output.

It also shows how to find a free TCP port with Socket, and notes that the whole 127.0.0.0/8 loopback range can be used when a fixed port is required.

Reading notes

#

• In tests, data generators help when creating examples by hand becomes tedious and the data ends up unrealistic.
• The text presents Faker as an option that produces data in English and more “human” data, without focusing on edge cases.
• Bogus.NET is shown as a broader library, with test object factories, support for multiple locales, and additional data sets.
• When using random data in tests, the text recommends controlling the seed so that a failure can be reproduced with the same data set.
• The seed can be set globally in Bogus.Randomizer.Seed or per instance with UseSeed.
• This style of testing is described as a simple form of property-based testing, because it generates random data and checks whether a system property holds.
• Another technique is to save a good reference of the result and compare it with the current output, which works well for rendering code such as HTML, images, or SVG.
• When the test fails in this model, the text suggests inspecting the differences and, if the change is expected, copying the files from current to reference and committing.
• For integration tests, the text shows how to discover a free TCP port using Socket bound to 127.0.0.1:0.
• If a fixed port is required, the text says the 127.0.0.0/8 range belongs to the local machine and allows using many loopback IPs, such as 127.0.0.2 and 127.0.0.3, on the same port.