Skip to content

make check in 1.8.1.

2 messages · Rolf Turner, Don MacQueen

Rolf Turner
# 
I just (finally!!!) got R version 1.8.1 to configure and build under
Solaris 9 (after much travail; there were funnies in my environment
variables that mucked things up, but that's another story).

Anyhow, when I ran ``make check'' I got an error right toward the
end.  Looking in the directory ``tests'' I found that the error was
associated with the file reg-tests-3.R, and the complaint was that it
couldn't find the object Cars93 from the MASS package.

So I edited reg-tests-3.R and stuck in the line ``data(Cars93)'':

===+===+===+===+===+===+===+===+===+===+===+===+===+===+===+===+===+===+===
## lm.influence where hat[1] == 1
if(require(MASS)) {
#
     data(Cars93)  # <------------------------- INSERTED LINE
#
    fit <- lm(formula = 1000/MPG.city ~ Weight + Cylinders + Type + EngineSize + DriveTrain, data = Cars93)
    print(lm.influence(fit))
    ## row 57 should have hat = 1 and resid=0.
    summary(influence.measures(fit))
}
## only last two cols in row 57 should be influential
===+===+===+===+===+===+===+===+===+===+===+===+===+===+===+===+===+===+===

Then I re-ran make check, and everything seemed to go OK.

I had a brief scan of the r-help mailing list archives just now and
could find no allusions to this problem.

Has anyone else encountered the problem?  ***Should*** that line
``data(Cars93)'' be in the code?  If not, what is going on?  If so,
why hasn't anyone else been bitten?  (Like, I mean, why me?) :-)


				cheers,

					Rolf Turner
					rolf at math.unb.ca

P. S.  Version details:
 > version
         _
platform sparc-sun-solaris2.9
arch     sparc
os       solaris2.9
system   sparc, solaris2.9
status
major    1
minor    8.1
year     2003
month    11
day      21
language R
Don MacQueen
# 
Whatever it is, it's apparently a bit on the subtle side. Perhaps a 
previous test loads the data, and for some reason that previous test 
doesn't run for you. I couldn't find one that did, though.

 From my 'make check':

running code in '../../source/tests/reg-tests-3.R' ... OK
comparing 'reg-tests-3.Rout' to 
'../../source/tests/reg-tests-3.Rout.save' ... OK

And my reg-tests-3.R does not have the data() statement:

if(require(MASS)) {
     fit <- lm(formula = 1000/MPG.city ~ Weight + Cylinders + Type + 
EngineSize + DriveTrain, data = Cars93)
     print(lm.influence(fit))
     ## row 57 should have hat = 1 and 
resid=0.                                                                             
     summary(influence.measures(fit))
}

-Don
At 4:49 PM -0400 2/10/04, Rolf Turner wrote: