We report activity data analysis on several open source software projects, focusing on time between modifications and on the number of files modified at once. Both have fat-tailed distributions, long-term memory, and display systematic non-trivial cr
oss-correlations, suggesting that quiet periods are followed by cascading modifications. In addition the maturity of a software project can be measured from the exponent of the distribution of inter-modification time. Finally, the dynamics of a single file displays ageing, the average rate of modifications decaying as a function of time following a power-law.