[Rd] A new way for an Alpha to fail

Douglas Bates bates@stat.wisc.edu
06 Mar 2000 08:47:54 -0600


Previously I had managed to get R-0.99.0 to compile and pass the tests
on a Debian Linux/Alpha 2.2 system by recompiling src/main/optim.c by
hand omitting the -O2 flag.

I just tried R-1.0.0 on a similarly configured machine, although not
the same machine as before, and it compiled ok but failed the tests.
This time it fails on
 bessel> x <- seq(0, 4, len = 501)

 bessel> plot(x, x, ylim = c(0, 6), ylab = "", type = "n", 
     main = "Bessel Functions  I_nu(x)")
 Error in axis(side, at, labels, ...) : no locations are finite

Doing some checking by hand shows that
 plot(x, x)
works and that range(x) is 0 to 4, as expected, but
 plot(x, x, ylim = c(0, 6))
fails with the above message.

The compilers are gcc 2.95.2 and g77 from 1999/10/xx

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