[Gretl-devel] Problem with dummify - extra

There is a further point that emerges when I experiment with dummify a bit more. The context is that I have a panel dataset with 91 panel units identified by the variable. However, for testing I have used the commands smpl id < 10 --restrict series pwork = pun_mean(&workers, &id) function pun_mean(series *x, series *id) discrete id list plist=dummify(x) print plist ... What I find is that dummify creates dummy variables for id=2 to id=91, not just the values id=2 to id=9 present in the sample supplied to the function. Surely, this cannot be the intention? Gordon