Mikito Harakiri wrote:
> The major problem with connect-by is that it doesn't seem to differentiate
> between nodes and edges of the graph, which is very confusing for the end
> user. For a graph being a tree the matter seems simple, as we can implicitly
> associate a node with the adjacent edge (although, there is an confusing
> ambiguity of 2 choices here as well). The second problem is that it is
> difficult to cost connect-by access method. Basically if you sees the
> execution plan with connect-by, then you can ignore the cost:-)
>
> I'm awaiting SQL Server Yukon release (I assume there is free developers
> version:-) to get my hands on "the latest and greatest" "recursive with"
> method...
I assume that your cost argument applies to "recursive with" equally
well. What kind of statistics/dynamic sampling can predict a number of
joins one have to perform recursively?
Received on Wed Jan 21 2004 - 20:32:11 CST
