UWN

#302 would help to begin with.

``` ?- phrase(("","";""),X). X = [] ; X = []. ?- setof(X,phrase(("","";""),X),Xs). Xs = [[]] ; Xs = [[]], unexpected. ?- setof(X,phrase(("";""),X),Xs). Xs = [[]]. ``` So this clearly is...

Currently, there is a cheap hack to circumvent goal expansion: ``` ?- setof(X,t^phrase(("","";""),X),Xs). Xs = [[]].

The expansion in the [cited source](https://www.complang.tuwien.ac.at/ulrich/Prolog-inedit/sicstus) (happens to) work for SICStus. The [variation for SWI](https://www.complang.tuwien.ac.at/ulrich/Prolog-inedit/swi/reif.pl) works for the examples in the paper. It fails to expand more complex examples. This...

There are a lot of unclear things related. Like: - the role of goal expansion as such. Does it have to be equivalent? Equivalent in what sense? Some years ago...

Further, - the update semantics of static code. In ISO, we have just the notion of preparation for execution, but there is no indication how to update static code. SICStus...

As for conformity there is an extension mechanism (5.5, 5.1 e). The problem behind is that the update semantics of SICStus prevents many optimizations that should remain user transparent (like...

(sorry for the delay) The version for [SICStus](https://www.complang.tuwien.ac.at/ulrich/Prolog-inedit/sicstus/reif.pl) and the current Scryer version differ in the way they handle erroneous cases, the cut other issues related to term-to-body conversion. In...

What kind of drop in performance. The idea would be that regular goals are not affected by this at all.

Goals that do not pose any problem (and thus where no instance would pose a problem) should really have no `call`-wrapper around.