> For GIOP users (and others I am sure), it would be nice to
> be able to have searchable fields in subtrees and sub-subtrees
> etc.
>
> This is especially important in GIOP as I want to preserve
> the data structure layout (eg: unions within structs). I do not
> want to have to flatten out all the data structures to make
> them searchable. I saw an old post from Gilbert on such matters
>
> <qoute>
> Because of the way the display filter routines work,
> a protocol that is placed in a subtree is not searchable.
> This is because the display filter routines expect protocols
> to be at the "top level" of the protocol tree. I could fix this,
> or I could keep this as a feature. For example, in this
> </quote>
He said "a protocol that is placed in a subtree is not searchable", not
"a field that is placed in a subtree is not searchable"; unless the
fields you're putting into the subtree are protocols (registered with
"proto_register_protocol()"), rather than regular fields (registered
with "proto_register_field_array()"), there's not an issue with the
level at which it's put in the protocol tree.
A "subtree" is anything under the top level; if nothing in a subtree
were searchable, you couldn't search for anything other than protocols,
as the *only* items that can validly appear at the top level of the
protocol tree are protocols, so all regular fields have to be in
subtrees. Given that fields *are* searchable, you clearly can put
filterable fields in subtrees....
