I recently wanted to use FitNesse for regression testing of a set of data provided by the customer, namely, about 200 models for a rich client application with some very poorly written XML serialization code. I wanted to use the Visual Studio's XSD.EXE utility to generate the C# for all the classes, and then verify that serialization and deserialization worked as expected. I could not even be sure that the schema I inherited was complete, so I wanted to read each file using a validation XML parser, then reserialize the resulting object graph, read it in
again and do some round-trip comparisons. Moreover, not all the models were still valid, so they might disappear from the directory, or new models might be dropped there, etc. To make a long story short, what I felt I needed was to be able to model a column fixture, with a "*" where the file name would appear, then generate the rows based on what was now in the directory. The markup would appear something like this:
!|Validation.BaseFileListingFixture|C:\FitNesse\dotnet|*.*|
|file name|do stuff?|
|*|0|
...and in this case the DoStuff() method would simply compare the file size to 0. Something like that.
The code works out differently in Java and C# (the former was easier; the latter required me to hack into Fit more than I would have liked). I can send you either one if you drop me a line. I had to eliminate the code from this posting because it ruined the format of the blog; can't have that!