[Rd] Re: [R] small error in regression tests

Martin Maechler maechler at stat.math.ethz.ch
Fri Mar 14 09:22:17 MET 2003


Thank you for the report.

[but this does *not* belong to R-help: it's about the "unstable"
 R development version!] 

>>>>> "AndyJ" == Andy Jaworski <apjaworski at mmm.com>
>>>>>     on Thu, 13 Mar 2003 17:00:00 -0600 writes:

    AndyJ> I just r-sync'ed the r-devel version of R-1.7.0
    AndyJ> (2003-03-11), compiled it under RH-8.0 and ran make
    AndyJ> check.  The reg-tests-2 fails at the very end with
    AndyJ> the message stating that "object cement was not
    AndyJ> found".  It looks like this piece of the regression
    AndyJ> test is new to version 1.7.  The the piece of code
    AndyJ> generating the error is:

    AndyJ> if(require(MASS)) {
    AndyJ> teststep <- function(formula, data)
    AndyJ> {
    AndyJ> d2 <- data
    AndyJ> fit <- lm(formula, data=d2)
    AndyJ> step(fit)
    AndyJ> }
    AndyJ> teststep(formula(y ~ .), cement)
    AndyJ> detach("package:MASS")
    AndyJ> }

    AndyJ> It seems that it requires
    AndyJ> data(cement)
    AndyJ> statement before the call to teststep.

Not quite.  It requires the corresponding version of the VR
bundle (with its main part MASS).

After unpacking the R-devel sources,
you need to get the **corresponding**  recommended packages by
typing
	tools/rsync-recommended

before you configure and make !
--
Regards,
Martin



More information about the R-devel mailing list