Thank you to all who replied with helpful suggestions. I had to run off to
meetings and
talks for a bit so am now processing it all.
1. It turns out that the issue was not with coxme, but with bsdmatrix, a
package that
coxme calls. It just happens to have a function ismat() with the same general
purpose and
some similar variable names, which led me down the rabbit hole. That package
contained a
"class(x) == " flaw, now fixed. (The fact that bdsmatrix has been stable and
unchanged
for nearly a decade helped with the deception.)
2. As pointed out by Duncan and Kurt, the coxme function also had a class(x)==
flaw. None
of my test cases triggered this, but since 'x' is an argument that can be
supplied by a
user, it certainly would have happened in package use. Good catch.
3. Dirk gave good input about the flags in R CMD check and how to find them.
One more
line in the "Writing R Extensions" manual would have been helpful, namely that
many of the
options are NOT available in the options() command nor as arguments to R. As
near as I
can tell, there is no way to turn on these logic checks within a standard R
session. A
desire to do this is where I started: I would have set options(warn=2,
error=recover) and
found the actual offender in a few minutes; and never had to bother all you
worthy readers.
4. I agree completely with Martin that errors like this should not be ignored.
In fact,
except for "variable may be used before initialized" messages from the C
compiler, I have
become grateful for EVERY complaint that comes out R CMD check. Notice the
verb "have
become" -- I did not start out so enthusiastic.
Again, thanks for the help.
Terry T.
[[alternative HTML version deleted]]
__
R-devel@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-devel