I entered the room of the Flow Visualization session barely in time to hear someone ask Christoph Garth to publish his code. YES! Only one session I have been on has not had at least someone ask this. If authors are not required to publish the code from which reviewers can check the assertions, we should at least make them think about doing so, by asking them at every session 🙂
And lest you think enforcing the publishing of code with papers is unfeasible, it has been tried before. SIGMOD 2008 actually had repeatability requirements (FAQ) which included submitting experimental evidence in the form of code, executables, and dataset.
So, ask for code!