all.equal
Paul Gilbert
pgilbert@bank-banque-canada.ca
Fri, 26 Jun 1998 14:12:32 -0400
>I don't have time to look at your new version,
>however, I *did* look at the last one and tried to replace
>However, your version wasn't good enough to "work" for 'make tests'
>[[ even after fixing the default tolerance which should be
> changed from
> .Machine$double.eps
>to sqrt(.Machine$double.eps)
Martin
The version I just posted is the same as the last one (called cut three). I'm
not sure what 'make tests' is suppose to do. I just ran it and it gave me a
bunch of error messages. It also looks like it may have changed some things to
mess up my working copy of R, but I'm not sure.
If you can be more specific about how this all.equal is not good enough for
'make tests' then I will try to fix it. I'm not sure what the general wisdom is
and I realize that .Machine$double.eps is a fairly tight comparison, but I
routinely test things to tighter tolerances than sqrt(.Machine$double.eps). The
change in C libraries between Splus 3.1 and 3.2 made large cumulative
differences in some of my simulations and I never would have found the problem
if I had only been checking calculations to sqrt(.Machine$double.eps).
Paul
-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
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
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._