[Rd] crash using fisher.test on Windows; was [R] unexpected crash of R under (PR#644)

[ligges@statistik.uni-dortmund.de]

Prof Brian D Ripley wrote:
> 
> On Thu, 24 Aug 2000, Emmanuel Paradis wrote:
> 
> > Dear All,
> >
> > We encountered an unexpected crash with R under Windows when running
> > fisher.test(). We had this problem only under this OS and only with the
> > following data:
> >
> > > T
> >       [,1] [,2]
> >  [1,]    2    1
> >  [2,]    2    1
> >  [3,]    4    0
> >  [4,]    8    0
> >  [5,]    6    0
> >  [6,]    0    0
> >  [7,]    1    0
> >  [8,]    1    1
> >  [9,]    7    1
> > [10,]    8    2
> > [11,]    1    0
> > [12,]    3    1
> > [13,]    1    1
> > [14,]    3    0
> > [15,]    7    2
> > [16,]    4    1
> > [17,]    2    0
> > [18,]    2    0
> > [19,]    2    0
> >
> > The commands:
> >
> > library(ctest)
> > fisher.test(T)
> >
> > result in a crash of R with a system error. (Note that if we change T[19,2]
> > to the value "1", then there is no problem.) The crash occurs with R-1.0.1
> > under Win98, and with R-1.1.1 under WinNT (both with Rgui and with Rterm).
> > However, there is no problem at all under Solaris (tested with R-1.0.1 and
> > R-1.1.0).
> 
> Your pinpointing the blame on the Windows port of R is quite unfair. The
> problem is in your usage of the function: the workspace was too small and
> the code was writing outside it.  It is pure chance that did not affect any
> the other OSes you tried (or was that just one OS?).
> 
> fisher.ctest(T, workspace=1e6) works without crashing on Windows, Solaris
> and Linux here.


I tried with the default of workspace=200000 on WinNT and Solaris. The
crash (Dr. Watson) only happens on WinNT. On Solaris everything works
fine.

Now I tried it using workspace=1000 (too small). On Solaris AND WinNT
the same result:

 > fisher.test(T, workspace=1000)
 Error in fisher.test(T, workspace = 1000) : 
         FEXACT error 40.
 Out of workspace.

This is the error message, I would expect, but not a crash of R!
CC'ing to R-Bugs with modified subject.


Uwe Ligges

-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
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
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._