[Cuis] Brainstorming question: what non-trivial uses can you think of for an object-based parser? (strings not invited)

[Phil (list)]

OK, I know we've been going on and on about OMeta recently but this one
has a Smalltalk-centric twist:  it really just struck me that sure,
OMeta can parse strings like nobody's business, but one of its claims to
fame is that it is actually a fully object oriented *object* parser.

So forgetting about all applications involving strings, what real world
uses can you think of where an object parser might come in handy?  You
have all of the capabilities that a parser brings to bear in terms of
pattern matching and being able to speculatively introspect objects in
its search, but on any arbitrary collection or stream of objects.  Keep
in mind that anything coming in over the wire (i.e. network) has already
been serialized so reconstituting it as an object graph seems
inefficient and of questionable value if the sole purpose in doing so is
to perform a task you could have easily done in it's default text/binary
form... but maybe I'm looking at it wrong.

The only thing that really jumps out at me is locally generated
time-series data (be it midi, audio, video, whatever) where you are
trying to prototype a solution without an existing framework and/or
experiment with with what the framework you are developing should look
like?  Another possibility (and I'm reaching here) is maybe some sort of
funky resource caching/scheduling scheme?  Would the value just be at
the investigative/prototyping stage or can you think of reasons why it
would make sense to stick with a parser-based approach longer term?