The segfault was occurring in Rf_asCharacterFactor, which was using
the levels attribute without sanity checks. Sanity checks are now
added (r68119 in trunk and r68120. in R-3-2-branch), which now gives
data(housing, package ="MASS")
x <- housing$Type + housing$Sat
Warning message:
Incompatible methods ("Ops.factor", "Ops.ordered") for "+"
match(x, unique(x))
Error in match(x, unique(x)) : malformed factor
I'll leave it to others to figure out how not to get the malformed
factor in the first place (but as a user could intentionally create
one we need the sanity checks anyway).
Best,
luke
On Mon, 30 Mar 2015, Hadley Wickham wrote:
I left out the warning - it's still there. The output object is
malformed but either +.factor should prevent this or match() should
check.
Hadley
On Mon, Mar 30, 2015 at 3:50 PM, William Dunlap <wdun...@tibco.com> wrote:
Did you leave out the warning from "+", which should be an error,
as it produces an illegal ordered factor in this case and factor+factor
is nonsensical? Or is the warning missing in the current development
version of R?
x <- factor("A", ordered=FALSE) + factor(c("B","C"), ordered=TRUE)
Warning message:
Incompatible methods ("Ops.factor", "Ops.ordered") for "+"
str(x) # 2 levels, so integer codes of 3 is illegal
ordered[1:2] w/ 2 levels B<C: 2 3
Bill Dunlap
TIBCO Software
wdunlap tibco.com
On Mon, Mar 30, 2015 at 1:10 PM, Hadley Wickham <h.wick...@gmail.com> wrote:
This is admittedly a contrived example, but...
data(housing, package ="MASS")
x <- housing$Type + housing$Sat
match(x, unique(x))
Hadley
--
http://had.co.nz/
______________________________________________
R-devel@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-devel
--
Luke Tierney
Ralph E. Wareham Professor of Mathematical Sciences
University of Iowa Phone: 319-335-3386
Department of Statistics and Fax: 319-335-3017
Actuarial Science
241 Schaeffer Hall email: luke-tier...@uiowa.edu
Iowa City, IA 52242 WWW: http://www.stat.uiowa.edu
______________________________________________
R-devel@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-devel
