Here is a little test:
:- begin_tests(tuturuu).
test(foo) :- Expected=[a,b],Got=[a,b,c],verdict(foo,Expected,Got).
% helper: transform lists A,B into ordered sets As,Bs
setify(A,B,As,Bs) :- % A,B -> As,Bs
list_to_ord_set(A,As), % A->As
list_to_ord_set(B,Bs). % B->Bs
% did the test pass? set-compare Expected and Got
verdict(Value,Expected,Got) :- % V,E,G -> b?
setify(Expected,Got,Eos,Gos),
format("For ~w, expected ~w and got ~w\n",[Value,Eos,Gos]),
ord_seteq(Eos,Gos)
-> true ; (format("For ~w, expected ~w but got ~w\n",[Value,Eos,Gos]),fail).
:- end_tests(tuturuu).
rt :- run_tests(tuturuu).
When I load this into SWI-Prolog:
?- [foo].
Warning: /home/somone/foo.pl:13:
Warning: Singleton variable in branch: Eos
Warning: Singleton variable in branch: Gos
true.
The ->
branch doesn't have access to the (local) variables Eos
and Gos
? But why?
It has access to Value
because that one appears in the head I suppose.
And indeed:
?- rt.
% PL-Unit: tuturuu For foo, expected [a,b] and got [a,b,c]
For foo, expected _29026 but got _29032 <--- YUP NO ACCESS
ERROR: /home/someone/foo.pl:3:
test foo: failed
done
% 1 test failed
% 0 tests passed
false.
It don't really see why this restriction occurs, expect that maybe it's a part that hasn't been implemented in the compiler?
(Btw, we need one of those "keep the bar in the green windows" for unit testing from the Java World, like this: (stolen from here)
Missing parenthesis? Try:
verdict(Value,Expected,Got) :- % V,E,G -> b?
setify(Expected,Got,Eos,Gos),
( ord_seteq(Eos,Gos)
-> true
; format("For ~w, expected ~w but got ~w\n",[Value,Eos,Gos]),
fail
).
Note that the standard operator definitions are:
?- current_op(Priority, Type, ',').
Priority = 1000,
Type = xfy.
?- current_op(Priority, Type, '->').
Priority = 1050,
Type = xfy.
W.r.t. "keep the bar in the green windows", you can do it easily with Logtalk lgtunit
tool (which you can also use to test Prolog code). E.g. https://logtalk.org/files/blog/2019_11_06/xunit_report.html
See https://logtalk.org/2019/11/06/testing-multiple-implementations-of-a-protocol.html and https://logtalk.org/2019/12/02/generating-code-coverage-reports.html for an introduction.