This is a very interesting change for SIGMOD 2008. Starting this year, there is a requirement for authors to submit data sets and code for replicating the experiments. For details, see:
I will be very interested to see how much of a burden this will be, how many times people will cite the use of proprietary software, and so on. I expect some small problems for this year but hopefully we will get used quickly to this mode of operation and start working on a paper with reproducibility of experiments in mind.
In any case, this is a very interesting development and I expect many conferences to follow.