st: gllamm poisson, robust standard errors

[Spencer Moore <mooresp@queensu.ca>]

Rabe-Hesketh and Skrondal mention in their book "Multilevel and 
Longitudinal Modeling Using Stata" using gllamm with the robust 
option to handle overdispersion at level 1 in 2-level 
random-intercept Poisson model.
When I use the robust option with the 2-level Poisson (gllamm), I 
keep obtaining the Stata message "can't get correct 
log-likelihood...something went wrong in comprob3".
Can anyone tell me what might be going awry in obtaining the robust 
standard errors?  I have used several different datasets and outcomes 
with this option.
thanks,
Spencer

*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/