R-alpha: set.seed compatibility

Paul Gilbert pgilbert@bank-banque-canada.ca
Mon, 15 Sep 1997 14:44:24 -0400

The modification to set.seed that I was suggesting does not change its
default behavior, so software which runs in S still runs with my change.
The change means that set.seed can also be given a value for
.Random.seed. This would fail with the usual S version of set.seed. I
don't think we need to take compatibility to the extend of failing when
S fails.

Also, I don't think set.seed can really accomplish what I think of as
the true objective of compatibility: reproduce S random experiments with
R (possibly not even if the same random number generator is used). Even
Splus has not been completely successful in this respect. There were
some changes between 3.2 and 3.3 which affected reproducibility.

Paul Gilbert

r-devel mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To: r-devel-request@stat.math.ethz.ch