Should we support DISTINCT?

[jezzsantos]

Is there a cross-repository semantic for DISTINCT?
Should we support it?

I like this idea of supporting DISTINCT as something more like: GROUPBY results by a specified field, sorting by the SortBy field and and then taking the TOP 1 result for that group (one result per group value). Is this the semantic that we want?

[jezzsantos]

If this is what we want, then what is the equivalent function in SQLServer?

(as far as I know, when you GROUPBY in SQL you cannot returns other field values in the query - we would want to return all the field of the TOP 1 result that is selected)