R help - Apr 2010 - Arima online forecasting

I'd like to have an online horizon=k forecast with Arima, something like:

library(forecast)

air.model = Arima(AirPassengers[0:100], c(1,0,1),
seasonal=list(order=c(0,1,1), period=12))

for(i in 0:(length(AirPassengers)-100)) {
        air.model2 = Arima(AirPassengers[0:(100+i)], model=air.model)
        cat(forecast(air.model2, h=2)$mean, '\n')
}

But this approach, even if not fitting the model from scratch, is still very
slow for big timeseries.

Some hints to speed it up?

Thanks,
Matteo

	[[alternative HTML version deleted]]