ARDL Model in eviews 9- stationarity

[spursinn]

I have a question about how the ARDL model works on Eviews 9 on the following aspect:
I see that in versions before eviews 9, for the ARDL model one has to specify the integration order of the variables, so that they are stationary. In this version I see people apply unit root test with break point, and, for example, two of the three series are I(1) and one is I(0). So next, when they select quick >> estimate equation >> ARDL... in the equation specification the variables that were found to be non-stationary are set without detrending or differencing, why?
So for example, gdp is non-stationary (it´s I(1) ), and in the equation specification one write L(gdp) instead of d(L(gdp)), then why without the differents?

Thak you for your help.

Regards.

[ajmair78]

Your question is reasonable.
Because it is ARDL not OLS. ARDL requires I(0) and I(1) variables. ARDL manages itself. I think so.
But OLS requires to write differenced variable.