Maths on values of properties that may not exist - gremlin

Question

Here is a query (gremlin-python; tinkerpop v3.3.3) that inserts a node with properties 'positive' and 'negative', then subtracts one from the other and outputs the result:
g.withSack(0).addV('test').property('positive', 2).property('negative', 3) \
.sack(assign).by('positive') \
.sack(minus).by('negative') \
.sack().next()
Out[13]: -1
Trying the same query except with one of those properties missing produces an error:
g.withSack(0).addV('test').property('positive', 2) \
.sack(assign).by('positive') \
.sack(minus).by('negative') \
.sack().next()
Out[13]: ... GremlinServerError: 500: The property does not exist as the key has no associated value for the provided element: v[36]:negative
I can get around this by coalescing the four possible cases:
g.withSack(0).addV('test').property('positive', 2) \
.coalesce(has('positive').has('negative').sack(assign).by('positive').sack(minus).by('negative').sack(), \
has('positive').values('positive'), \
has('negative').sack(minus).by('negative').sack(), \
sack() \
).next()
It sure is ugly though - is there a neater solution? Ideally there would be the possibility to insert a default value in the absence of a property. I've tried using the 'math' step as well, but it's only slightly neater and doesn't avoid the problem of non-existent properties. To be clear, In the case of multiple traversers, I want a result for each traverser.

Answer 1

I think if you do math() or sack() to solve this, you should probably consider going with the idea of having "required" properties on these vertices on which you intend to do these calculations. That should make things much easier. I do feel like math() would be neater, though you said otherwise:
g.V().as('a').out('knows').as('b').
math("a - b").
by(coalesce(values('hasSomeValue'), constant(0))).
by(coalesce(values('missingValue'), constant(0)))
That's pretty straightforward though perhaps your examples were meant for simplicity and you have a lot more complexity to consider.
I suppose Gremlin could be changed to allow for a second parameter in the by() as some default if the first traversal did not return anything, thus:
g.V().as('a').out('knows').as('b').
math("a - b").
by(values('hasSomeValue'), constant(0)).
by(values('missingValue'), constant(0))
Saves some typing I'd suppose, but I'm not sure that it is as clear to read as with the use of coalesce(). I think I like the explicit use of coalesce() better.