So I have taken it that you want to order by attribute name (rather than attribute value) and that this is purely something you want from the user interface because you have lots and lots of attribute names.
There may be better ways, but I just create a 4 step process which involves reading the schema and create an ordered schema, which can be imported via data cache into the attribute manager. You could of course just write that feature out into a separate file (like excel) and just import that into the attributemanager.
Then you can connect your real data connection, and remove all the processes used to sort the schema.
There may be a better ways, but this certainly works.
I have ordered each step with 1 through to 4. Hopefully the basic instruction makes sense.