panel estimation options

[eng2010]

In the Least Squares Panel Options, there are three boxes to choose. The first box is Effects specification where we can choose fixed or random effects in the cross-section or period. The second box is Weights option where we can choose different kinds of GLS weights. The third box is Coef covariance method.

My understanding is that once you choose the second box, you do not need to choose the third box. For example, suppose you choose the GLS time period SUR (one kind of Park's estimator), the GLS estimator will be used to correct the period heteroscedasticity and period correlation. But if you choose the PCSE time period SUR (in the third box), an OLS estimator with PCSE will be used for estimation. Please confirm if this is exactly what Eviews does in the estimation?

But I found out that when I choose both the second box with GLS time period SUR and the third box with PCSE time period SUR, Eviews will report an estimation result. So my second question is why Eviews can use the two different estimators at the same time for estimation? Conceptually, you use either "GLS time period SUR" or "OLS with PCSE" to correct the period heteroscedasticity and period correlation. But you cannot use two estimators at the same time? Is something wrong with Eviews program? Please advise.

Thanks,
Eric