Skip to content

Which.min bug?

3 messages · Elliot Williams, Martin Maechler

Original
Elliot Williams
# 
Hi,

I'm not sure this is a bug, so I thought I'd bounce it off the help group 
first.  

I had a dataset which I was subsetting, and occasionally I get an empty 
subset.  If I don't check for emptiness and go straight to a which.min call 
on the subset, the program gets a big negative number back.

One-line Example:
[1] -2147483647

This caused an indexing error which was hard(ish) to track down.  Should it 
just return NA or throw an error?  Should I re-submit to r-bugs?  Or is there 
some deep reason for this behavior that I don't understand?

Thanks.
			Elliot.
10 days later
Martin Maechler
# 
Elliot> Hi, I'm not sure this is a bug, so I thought I'd bounce it off
    Elliot> the help group first.

    Elliot> I had a dataset which I was subsetting, and occasionally I get
    Elliot> an empty subset.  If I don't check for emptiness and go
    Elliot> straight to a which.min call on the subset, the program gets a
    Elliot> big negative number back.

    Elliot> One-line Example:
  
    >> which.max(NULL)
    Elliot> [1] -2147483647

(this happens in C code, by   "NA_INTEGER + 1"
 which unfortunately is *not* NA ..)

    Elliot> This caused an indexing error which was hard(ish) to track
    Elliot> down.  Should it just return NA or throw an error?  Should I
    Elliot> re-submit to r-bugs?  Or is there some deep reason for this
    Elliot> behavior that I don't understand?

Yes it is a bug (and I am the culprit), and the same for  which.max().
As I have been away for almost two weeks, I'm sorry that nobody else has
confirmed this. 

I believe, for consistency

  which.min(v)
		should return the same as
  which(min(v) == v)

and hence should return  

  integer(0)

here.  Returning NA instead is another intuitive possibility.
What do others think?

Martin Maechler <maechler at stat.math.ethz.ch>	http://stat.ethz.ch/~maechler/
Seminar fuer Statistik, ETH-Zentrum  LEO D10	Leonhardstr. 27
ETH (Federal Inst. Technology)	8092 Zurich	SWITZERLAND
phone: x-41-1-632-3408		fax: ...-1228			<><
-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-help 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-help-request at stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._
Martin Maechler
# 
Elliot> Hi, I'm not sure this is a bug, so I thought I'd bounce it off
    Elliot> the help group first.

    Elliot> I had a dataset which I was subsetting, and occasionally I get
    Elliot> an empty subset.  If I don't check for emptiness and go
    Elliot> straight to a which.min call on the subset, the program gets a
    Elliot> big negative number back.

    Elliot> One-line Example:
  
    >>> which.max(NULL)
    Elliot> [1] -2147483647

    MM> (this happens in C code, by "NA_INTEGER + 1" which unfortunately is
    MM> *not* NA ..)

    Elliot> This caused an indexing error which was hard(ish) to track
    Elliot> down.  Should it just return NA or throw an error?  Should I
    Elliot> re-submit to r-bugs?  Or is there some deep reason for this
    Elliot> behavior that I don't understand?

    MM> Yes it is a bug (and I am the culprit), and the same for
    MM> which.max().  As I have been away for almost two weeks, I'm sorry
    MM> that nobody else has confirmed this.

    MM> I believe, for consistency

    MM>   which.min(v) should return the same as which(min(v) == v)

    MM> and hence should return

    MM>   integer(0)

which I have committed for "R-release", to be version 1.2.2 in a few days.

    MM> here.  Returning NA instead is another intuitive possibility.  What
    MM> do others think?

Martin Maechler <maechler at stat.math.ethz.ch>	http://stat.ethz.ch/~maechler/
Seminar fuer Statistik, ETH-Zentrum  LEO D10	Leonhardstr. 27
ETH (Federal Inst. Technology)	8092 Zurich	SWITZERLAND
phone: x-41-1-632-3408		fax: ...-1228			<><
-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-help 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-help-request at stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._