Some of the simple finalization examples could be easily converted into HUnit tests. We use a `Reader` monad with an `IORef` containing a list of "actions" like "open file", "close file" and "exception" to keep track of what happens during pipeline execution.
The finalizer examples are not in cabal, since they don't even have a main. I will probably include them as unit tests later.
Fixed and updated benchmarks to the latest conduit and pipes-extra. Benchmarks are now integrated in cabal, so you can run them with something like: cabal configure --enable-benchmarks cabal build cabal bench