>Why not test to see if all f1 are NULL first, if so make x NULL, otherwise calculate the average and assign it to x?
>
>>Thanks for the suggestion Nadya. The SQL result is what I'm looking for but unfortunately, I have to get the results from a buffered view and so it seems as if I'm no further ahead! Sigh....

It would mean a scan for the whole file :( Say, file has 500000 records and all f1 values are null. Is there a simple way to determine, if all records are null or not? I don't see a way except for scaning the whole file... :(