Simple bug in lmer: reliably crashes R

There seems to be a minor input-validation bug in lme4 0.99875-8,
where fitting a model with (0 | ...) crashes R.

To reproduce:

library(lme4)
frame <- data.frame(A=1:100, B=1:100, C=rep(1:10, 10)
lmer(A ~ B + (0 | C), frame)

Process R floating point exception at Sat Oct 13 01:26:25 2007

Obviously the workaround is "don't do that then", but it would
probably be good to fix at some point.

I have just uploaded lme4_0.99875-9 to CRAN's incoming directory.
This version generates an error when trying to fit a model with terms
that reduce to (0|grp).

Thanks for the report.