Encountered a very strange bug and can't quite figure out what's going on.

[blackwint3r]

In the following example, I aim to find all the "sub-graph" of a named graph, then merge its own value with the values of the sub graph and output the result.

@prefix : <http://www.example.com/ns#> .
@prefix e: <http://eulersharp.sourceforge.net/2003/03swap/log-rules#> .
@prefix log: <http://www.w3.org/2000/10/swap/log#> .

:g1 :sub-graph :g2.
:g1 :value {:a :b :c}.
:g2 :value {:bob :say {?x :can :fly}}. # This bug only occurs when `:g2` contains a graph, and within that graph, there exists a variable.
#:g2 :value {?bob :say {:x :can :fly}}. # Changing `:g2` to this works fine

# When parameter (`?gs`) is an empty list, the union of graphs should be an empty graph(i.e., true). However, `log:conjunction` will return false.
# Therefore, I additionally defined `:graph-union`.
{?gs :graph-union ?g} <= {?gs log:notEqualTo (). ?gs log:conjunction ?g.  }.
{?gs :graph-union true} <= {?gs log:equalTo (). }. 

# `:value2` merges the values of its subgraphs with its own value
{?g :value2 ?x } <= {
    ?g :value ?x1.
    [] e:findall (?x2 {?g :sub-graph ?g2. ?g2 :value2 ?x2} ?answer). 
    ?answer :graph-union ?x3.
    (?x3 ?x1) log:conjunction ?x.
}.

{:g1 :value2 ?x. "T0" log:trace  ?x.} => {:answer :is ?x}.

eye --pass-only-new --quiet test.n3 did not print the expected output( :answer :is ... ):

"T0" TRACE {
    :bob :say {
        ?_4986 :can :fly.
    }.
    :a :b :c.
}
@prefix : <http://www.example.com/ns#>.

What strikes me as most peculiar is that the log:trace statement could correctly print the debug message, yet this forward rule does not execute.
I also trace the results of the intermediate variables, and they were all correct. The only issue is that the final forward rule does not produce any output.

[josd]

The issue is that eye is not producing universal statements, that is triples with ? quickvars in it. The only exception is producing log:implies triples and in the latest eye when you use --no-ucall.
So now with EYE v11.12.11 (2025-03-21)

eye --quiet --nope --no-ucall test.n3 --pass-only-new

should give

"T0" TRACE {
    :bob :say {
        ?_8826 :can :fly.
    }.
    :a :b :c.
}
"T0" TRACE {
    :bob :say {
        ?_8980 :can :fly.
    }.
    :a :b :c.
}
@prefix : <http://www.example.com/ns#>.


:answer :is {
    :bob :say {
        _:sk_0 :can :fly.
    }.
    :a :b :c.
}.

[blackwint3r]

Yes, --no-ucall produces the expected output, solved my practical problems.

But I still have some confusion, is it expected behavior that --pass-only-new alone does not output triples containing ?quickvars?

However, the simple example:

:g3 :value2 {:a :b :c. :bob :say {?x :can :fly}}.
{:g3 :value2 ?x} => {:answer2 :is ?x}.

correctly outputs without using --no-ucall.

Additionally, you mentioned "The only exception is producing log:implies triples", but neither :g2 :value {@forAll :x. {:x a :bird} log:implies {:x :can :fly}}., nor :g2 :value {{?x a :bird} log:implies {?x :can :fly}}.in the subgraph example produces no output without using --no-ucall.

[josd]

You are right and we need a new switch

    --no-call-residue-vars          no check for residue vars for forward rules

It is implemented in EYE v11.12.12 (2025-03-21) and now you should get

eye --quiet --nope --no-call-residue-vars test.n3 --pass-only-new
"T0" TRACE {
    :bob :say {
        ?_8522 :can :fly.
    }.
    :a :b :c.
}
"T0" TRACE {
    :bob :say {
        ?_8676 :can :fly.
    }.
    :a :b :c.
}
@prefix : <http://www.example.com/ns#>.

:answer :is {
    :bob :say {
        _:sk_0 :can :fly.
    }.
    :a :b :c.
}.

Sorry for the confusion with --ucall which is now back to what it was.
(I had some fever so coudn't problably think that clearly)

[blackwint3r]

Thanks! Sorry to hear that you weren't feeling well. Hope you're recovering well, health comes first!

[josd]

Is fine now and FYI, we renamed log:program to log:pro to honor prolog 😉