top() in the middle of a KQL query

[alightgoesout]

The KQL grammar (on the v2 branch) allows the top() filter in the middle of a query. What should a > top() > b return ? I see several possibilities:

• return the same result as top() > b (after all, the spec says that top returns all toplevel children of the current document)
• return an empty set (the query does not make sense)
• return the same result as a > b (ignore top(), but that seems weird)
• fail the execution of the query (but I'd rather fail parsing instead)

[zkat]

If you look at the actual grammar on that branch, top() is only allowed at the very toplevel of a selector:

query-str := $bom? query
query := selector q-ws* "||" q-ws* query | selector
selector := filter q-ws* selector-operator q-ws* selector | filter
selector-operator := ">>" | ">" | "++" | "+"
filter := "top(" q-ws* ")" | matchers

https://github.com/kdl-org/kdl/blob/kdl-v2/QUERY-SPEC.md#full-grammar

Unless I'm misreading my own grammar/spec wrong. Maybe what's missing is mentioning this fact further up in the spec?

[alightgoesout]

You're right I misread the grammar, my bad.

[larsgw]

I'm not sure that's the case? selector can expand to filter q-ws* selector-operator q-ws* filter right? I don't think this should be a syntax error though, it should return an empty set IMO.

[zkat]

oh yeah, you're right.

I definitely don't think we should have top() outside of the toplevel, though. I don't think we should just default to a random empty value.