As regular readers know, I'm a big fan of Doug Bates's lmer() function, which comes with his lme package in R. Anyway, I had an exchange with another lmer() user which included the following funny bit:
Using the version from CRAN (lme 0.99875-9) everything works; however using lme4_0.999375-0 it does not! I was running lme4_0.999375-0...
I'm thinking that maybe it's time to go to version 1.0 already!
I think they're taking a limiting approach (just like latex is approaching pi in their version numbers). An irrational number would have been more interesting though ;)