This release has fixes for the various sasdata2dataframe methods (aliases to the one code base), and a new, second, implementation which uses an intermediary CSV file to improve performance for larger data sets. The original implementation, which is still the default, builds the dataframe in memory as the data is streamed from SAS. That method doesn't scale. Pandas performs better reading large data from disk and creating the dataframe. The original method was reworked so it will run without consuming significant system resources, like the first implementation ended up doing with larger data. saspy version 2.1.8 had this in it, but it also had a couple bugs, so 2.2.1 is basically 2.1.8 with the fixes for those couple bugs.