[Rd] make check fails d-p-q-r-tests.R - OpenBSD 3.0 (PR#1461)

maechler@stat.math.ethz.ch maechler@stat.math.ethz.ch
Mon, 15 Apr 2002 12:46:46 +0200 (MET DST)

>>>>> "Jason" == Jason Turner <jasont@indigoindustrial.co.nz> writes:

    Jason> Found one that stumped me.  I did a search on R-bugs for the
    Jason> string "d-p-q-r-tests", and found zero entries in all categories 
    Jason> (I don't think I secretly gave it a regex).

    Jason> Did the usual configure, make, make check on OpenBSD 3.0, and 
    Jason> encountered this message on d-p-q-r-tests.Rout:

    Jason> running code in `d-p-q-r-tests.R' ... OK
    Jason> comparing `d-p-q-r-tests.Rout' to `./d-p-q-r-tests.Rout.save' ...251,253c251
    Jason> < [1] "`is.NA' value mismatches: 800 in current, 0  in target"
    Jason> < Warning message:
    Jason> < NaNs produced in: exp(pab <- dbeta(p, a, b, log = TRUE))
    Jason> ---
    >> [1] TRUE
    Jason> 315,317c313
    Jason> < [1] "`is.NA' value mismatches: 4 in current, 2  in target"
    Jason> < Warning message:
    Jason> < NaNs produced in: exp(z)
    Jason> ---
    >> [1] TRUE
    Jason> OK

    Jason> If you're not too busy, I'd love to provide you with whatever you
    Jason> need to look at this more closely than my untrained eyes can.
    Jason> config.cache?  d-p-q-r-tests.Rout?  Anything else?

Actually, we'd be very grateful if you could run these with
R-devel aka "R pre1.5.0" instead.

You can get a snapshot using rsync (or the tar file), see

If you see the same problem (or also with 1.4.1)
look at the tests/d-p-q files
probably start R and try things like

 z <- c(-Inf,Inf,NA,NaN, rt(1000, df=2))
 z.ok <- z > -37.5 | !is.finite(z)
 for(df in 1:10)
    if(!is.logical(all.equal(pt(z, df), 1 - pt(-z,df), tol= 1e-15)))
      cat("ERROR -- df = ", df, "\n")

 c(pz <- pnorm(z), 1 - pnorm(z, lower=FALSE))
 c(pz, pnorm(-z, lower=FALSE))
 c(log(pz[z.ok]),  pnorm(z[z.ok], log=TRUE))

because these are around line 315 of tests/d-p-q-r-tests.Rout.save
and the second problem happened around there.

[A more desperate option would be to offer a login into your computer to
 someone from the R core team ...]

Thanks for helping us to have R running correctly on OpenBSD 3.0,

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