A very useful feature IMHO would be to be able to run crlmm analysis, both genotype and cnv on a subset of CEL files. It would allow to implement tests, and to run benchmarks much much faster.
It could be a probe_ids= or probeset_ids= extra arguments to genotype() and crlmm().
Some companion convenient functions such as fetch_probe_ids(by_chr=, ...) could leverage this new feature.
My use case is that I have unit tests to check my pipeline, and I need a minimum of CEL files otherwise crlmm complain, and it takes ages, so that I tends to undertest it.