[R] reproducibility of the loess function

[Marie-Agnès DILLIES]

Hi,

We noticed a difference between R versions 1.9.1 and 2.0.1 concerning the 
loess function. The resulting fitted data is quite different. Furthermore, 
with the 2.0.1 release, if I apply the loess function several times on the 
same input data set, I get different results.

Has anybody already noticed such a problem ?

Thanks

Marie-Agnès