[Rd] C function is wrong under Windows 7

[Thomas Lumley]

On Mon, Oct 24, 2011 at 6:04 AM, Evarist Planet
<evarist.planet at irbbarcelona.org> wrote:
> Dear mailing list,
>
> I have a C function that gives me a wrong result when I run it under Windows
> 7.

The fact that you extract pointers to the contents of fchr and sign
before coercing them to REALSXP is deeply suspicious to me, though I
admit I don't see how it causes the problem, since you don't use the
new versions, you don't overwrite them, and the old versions should
still be protected as arguments to the function.

> SEXP getEs(SEXP fchr, SEXP sign)
> {
>  int i, nfchr, nsign;
>  double *rfchr = REAL(fchr), *res;
>  int *rsign = INTEGER(sign);
>
>  PROTECT(fchr = coerceVector(fchr, REALSXP));
>  PROTECT(sign = coerceVector(sign, REALSXP));
>
>  nfchr = length(fchr);
>  nsign = length(sign);
>
>  SEXP es;
>  PROTECT(es = allocVector(REALSXP, nfchr));
>  res = REAL(es);
>
>  double nr = getNr(rfchr, rsign, nsign);
>
>  SEXP phit;
>  PROTECT(phit = allocVector(REALSXP, nfchr));
>  double *rphit;
>  rphit = REAL(phit);
>  for(i = 0; i < nfchr; i++) rphit[i] = 0.0;
>  getPhit(rfchr, rsign, nsign, nr, rphit);
>  cumsum(rphit, nfchr);
>
>  SEXP pmiss;
>  PROTECT(pmiss = allocVector(REALSXP, nfchr));
>  double *rpmiss;
>  rpmiss = REAL(pmiss);
>  for(i = 0; i < nfchr; i++) rpmiss[i] = 0.0;
>  getPmiss(rsign, nfchr, nsign, rpmiss);
>  cumsum(rpmiss, nfchr);
>
>  for (i = 0; i < nfchr; ++i) {
>   res[i] = rphit[i] - rpmiss[i];
>   }
>
>  UNPROTECT(5);
>  return es;
> }
>
> Could you please help me to find out what I am doing wrong?
>
> Many thanks in advance,
>
> --
> Evarist Planet
> Research officer, Bioinformatics and Biostatistics unit
> IRB Barcelona
> Tel (+34) 93 402 0553
> Fax (+34) 93 402 0257
>
> evarist.planet at irbbarcelona.org
> http://www.irbbarcelona.org/bioinformatics
>
>        [[alternative HTML version deleted]]
>
> ______________________________________________
> R-devel at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-devel
>



-- 
Thomas Lumley
Professor of Biostatistics
University of Auckland