Spurious choice-points in frame.pl ?

[pmoura]

Are the implicit choice-points created by the following member/2 and select/3 calls required?

https://github.com/terminusdb/terminus-server/blob/df1ba67477fd0d32f9f4062eb18e62e2982b2117/library/frame.pl#L818-L819

The call to the realise_triples/4 predicate from the object_edges/3 predicate suggests that those calls could be replaced with memberchk/2 and selectchk/3. As the realise_triples/4 predicate is recursive, these choice-points accumulate.

P.S. Given sort/2 semantics, the object_edges/3 predicate can also be rewritten as:

object_edges(URI,Database,Edges) :-
    (   most_specific_type(URI,Class,Database),
        class_frame(Class,Database,Frame),
        realise_triples(URI,Frame,Database,Unsorted)
    ->  sort(Unsorted,Edges)
    % There is no type in the database, so it doesn't exist...
    ;   Edges=[]).

[GavinMendelGleason]

The choice points are indeed spurious. This entire section should be rewriten and the object_edges version looks better as well.

[GavinMendelGleason]

Fixed in e28b1f4