Strange Error on AIX

Martin Maechler Martin Maechler <maechler@stat.math.ethz.ch>
Sat, 6 Jun 1998 15:08:37 +0200


>>>>> "Janusz" == Janusz Kawczak <jkawczak@fisher.stats.uwo.ca> writes:

[You wrote to "owner-r-help@...."  -- that's just me (MM);
 why did you not use the mailing list ? (R-devel or R-help)]

    Janusz> Hello, I am getting this 'strange' message when running the R
    Janusz> program on AIX 4.2.1. I have not had any problems with the
    Janusz> compilation of the program. Does anyone have an idea due to
    Janusz> what is the following error present:

    Janusz> R : Copyright 1998, Robert Gentleman and Ross Ihaka Version
    Janusz> 0.61.3 (May 3, 1998)

    >> plot(rnorm(100),rnorm(100))
    Janusz> Gpretty(NaNQ,NaNQ,5): cell=NaNQ, ndiv= -2 <=0;
    Janusz> (ns,nu)=(-2147483647,2147483647); dx=NaNQ, unit=NaNQ, ismall=0.
    Janusz> Gpretty(NaNQ,NaNQ,5): cell=NaNQ, ndiv= -2 <=0;
    Janusz> (ns,nu)=(-2147483647,2147483647); dx=NaNQ, unit=NaNQ, ismall=0.
    Janusz> Error: Length cannot be negative

    Janusz> In addition to the above, I've tried the latest snapshot
    Janusz> version of R; the same problem occurs.

Janusz, this *IS* a problem that I recently reported to  R-devel.
Ive also access to an AIX 4.2 (older than 4.2.1 ?) with the same problem:

I've tracked the bug to

	> min(1)
	NaNQ
	
(the same for max(.) for any number, and hence for range(.) ..)
The Gpretty(.) message is a direct consequence of this broken arithmetic.

Peter Dalgaard recommended using a debugger and seeing where in the source
code things go haywire..
I haven't had time yet...

Are you also using gcc for compilation?
Could it be a gcc--AIX misconfiguration?

If you can start debugging (e.g. R --debugger gdb )
we'd be glad if you worked with a recent snapshot (0.62), since quite a few
things have changed, notably the IEEE arithmetic (Inf, -Inf and NaN).

Martin
-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
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
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._