comparing two boolean values and the unary 'not'

[Pippin555]

This is not an issue, but a request for a smarter solution,

I would like to compare two boolean values with an 'is' operator. But that operator is not available for groovysh_evaluate.

    t = (g.V()
         .limit(1)
         .sack(assign)
         .by(constant(true))
         .sack(is)
         .by(constant(false))
         .sack()
         )

When I try to negate a boolean value, there is no operator for that either.

          .sack(assign)
           .by(constant(true))
           .sack(not)
          .sack()

In the end I wrote a solution with a .choose, comparing with constant(true), see below.

Another solution is to convert boolean to int 0 and 1 and then use .math('x - y'), when that gives 0, the booleans are the same.

I can't believe that there is no shorter way to achieve this.

This determines the exclusive-or for the values in as('x') and as('y'):

        .choose(
                or_(
                    and_(
                        select('x').is(true),
                        select('y').is(true)),
                    and_(
                        select('x').is(false),
                        select('y').is(false))),
                constant(false),
                constant(true))

[Pippin555]

Just a couple of chapters further on:

             .choose(where('x', neq('y')),
                     constant(true),
                     constant(false))

this solves the first issue:

x y xor F F F F T T T F T T T F

[krlawrence]

You could implement boolean not logic using a Coalesce step

gremlin> g.inject(false).coalesce(is(true).constant(false),is(false).constant(true))
==>true  

[krlawrence]

Also I updated the section that talks about choose and option over the weekend. The option step now accepts predicates which may also help in your case.

[Pippin555]

a small change to your example: 'GraphTraversalSource" object has no attribute 'coalesce', you'll need g.V()

My Python implementation:

    print('option')
    for x in [False, True]:
        t = (g.V()
             .choose(constant(x))
             .option(True, constant(False))
             .option(False, constant(True)))
        r = next(t, None)
        print(f'{x}, {r}')

[krlawrence]

In my example above the V() is not needed as the inject is there. That turns the source into a Traversal. Keep in mind that g.V() looks at ALL vertices in the graph unless constrained by a filter like has or where.

[Pippin555]

I used g.coalesce(constant(true).is(true).constant(false),constant(true)) and that does not work...

About V(): yes, I already have inadvertently doubled all vertices in the database when I used g.V().addV(...) , wondering where all those vertices came from...

on the side: Could you add in chapter 4.12 a word about where lambda functions are executed? I sent a question about those functions to the makers of ArangoDB Tinkerpop Provider and they asked me whether it would be executed "in the database" (I would say "on the server side") and they do not allow that.

[krlawrence]

Now that work has started on a second edition we will take the remaining discussions in this issue as part of that work.