> However, in the case of ML, I would stick with your original advice: use full 
> paths to elements rather than //. In ML this allows the processor to use 
> several indexes to resolve the full path, rather than a single index. 

The biggest win of using a full path instead of // in Mark Logic is that once 
the fragment is fetched, it allows the server to more quickly find the desired 
node within the fragment.  With // the server has to walk the whole fragment 
tree while with a direct path it can go straight to the right node.

With xdmp:estimate(), things can actually be faster by using // because that 
reduces the number of term lists that have to be intersected, and since you're 
not fetching fragments the advantage listed above doesn't apply.

Things are usually fast either way, but it's fun to maximize.

A beginners rule of thumb would be to use full paths when possible and 
appropriate for your query.

-jh-

_______________________________________________
General mailing list
[email protected]
http://xqzone.com/mailman/listinfo/general

Reply via email to