[R] Question about quantile.default

Ben Bolker bolker at ufl.edu
Sat Oct 4 23:59:57 CEST 2008

Peter Dalgaard <p.dalgaard <at> biostat.ku.dk> writes:

[snip discussion of overrun in seq()] 

> >   If I don't hear otherwise I will submit this as
> > a bug to r-devel  ...
> >  
> It IS deliberate.... We discussed this when R was still in the toddler 
> stages.
> If you don't allow slight overruns like that, you end up with sequences 
> sometimes being one element short, so seq() has a small "do what I mean" 
> fudge factor built in. We might polish off the end value though, or fix 
> quantile() to be less hysterical about values slightly out of range.

  Polishing the end value seems to be a good idea.  If not,
shouldn't one at least change the documentation to indicate
that the upper value is not absolute -- i.e. that such overruns
can occur?

    Ben Bolker

More information about the R-help mailing list