R help - Aug 2007 - svyglm.predict: cannot allocate vector

Hi,

I have a problem with the predict function of svyglm:
When I try to predict with an object of svyglm I get
the error message that R can't allocate a vector about
3 GB big.

Has anybody a hint how I can reduce the memory used
for this?

I have about 25000 observation and use the weigths
parameter. Where is so much data created?

Thanks,
  Werner