Interaction of "not" and "match" steps

[lists+gre...@krejci.pw]

Hi,

In the query below, I'm trying to find all vertices that have "propA" and
"propB" properties set to a different value. The below query complains about
an unsolvable match pattern. Is that a bug or am I doing something wrong in
the query?

graph.addVertex("propertyA", 1, "propertyB", 2);

graph.traversal().V().not(
__.match(
__.as("a").values("propertyA").as("propA"),
__.as("a").values("propertyB").as("propB"))
.where("propA", P.eq("propB"))
.select("a")
);

Because the above is a (simplified) generated query, I'm using a "not" step
instead of simply using P.neq.

Regards,

--
Lukas Krejci

[Daniel Kuppitz]

Why so complicated? This should provide the correct result:

g.V().match(

    __.as("a").values("propertyA").as("propA"),

    __.as("a").values("propertyB").as("propB")

  ).where("propA", neq("propB")).select("a")

Another way would be:

g.V().filter(valueMap("propertyA","propertyB").where("propertyA", neq("propertyB")))

Cheers,

Daniel

--
Lukas Krejci

--
You received this message because you are subscribed to the Google Groups "Gremlin-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to gremlin-users+unsubscribe@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/gremlin-users/50896973.2DW1h1rdbW%40localhost.localdomain.
For more options, visit https://groups.google.com/d/optout.

[lists+gre...@krejci.pw]

As I said, we generate the query based on inputs. The "match" part needs to
sometimes be negated, sometimes not. So my initial approach to this was to
simply leave the query "snippet" as is and just negate it when needed. That's
when I discovered this problem.

While in this simplified example P.eq -> P.neq is indeed the simplest
solution, bringing it back to our codebase would be more difficult, because
the "real" query is more complex. I simplified the query to illustrate the
strange behavior...

email to gremlin-user...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/
gremlin-users/50896973.2DW1h1rdbW%40localhost.localdomain.
For more options, visit https://groups.google.com/d/optout.

--
You received this message because you are subscribed to the Google Groups
"Gremlin-users" group.
To unsubscribe from this group and stop receiving emails from it, send an

email to gremlin-user...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/

gremlin-users/CA%2Bf9seWqOM%3DXCT0nXuZ%3DfTu3neW_3uiZCsEAQmeHLBqfsB4baw
%40mail.gmail.com.

For more options, visit https://groups.google.com/d/optout.

--
Lukas Krejci

[Daniel Kuppitz]

Looks like you've found a bug in NotStep. Here are 2 ugly workarounds:

1. Enable path computations:

g.withPath().V().not(match(...))

2. Add requirements in the parent traversal:

g.V().as("a").not(match(...)).select("a")

Cheers,

Daniel

email to gremlin-users+unsubscribe@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/
gremlin-users/50896973.2DW1h1rdbW%40localhost.localdomain.
For more options, visit https://groups.google.com/d/optout.



--
You received this message because you are subscribed to the Google Groups
"Gremlin-users" group.
To unsubscribe from this group and stop receiving emails from it, send an

email to gremlin-users+unsubscribe@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/

gremlin-users/CA%2Bf9seWqOM%3DXCT0nXuZ%3DfTu3neW_3uiZCsEAQmeHLBqfsB4baw
%40mail.gmail.com.
For more options, visit https://groups.google.com/d/optout.

--
Lukas Krejci

--
You received this message because you are subscribed to the Google Groups "Gremlin-users" group.

To unsubscribe from this group and stop receiving emails from it, send an email to gremlin-users+unsubscribe@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/gremlin-users/3581542.WupXtlzBPQ%40localhost.localdomain.

[Robert Dale]

I was thinking that the original query had a misplaced not. It
doesn't make sense to me to not() a select(). What would be the
expected outcome? That is, what is the opposite of a vertex?

I agree what you found is bug. not(match()) should work.

These should also be valid not() queries:

1. not(where()):
g.V().match(__.as("a").values("propertyA").as("propA"),__.as("a").values("propertyB").as("propB")).not(where("propA",
P.eq("propB"))).select("a")

2. where(not()):
g.V().match(__.as("a").values("propertyA").as("propA"),__.as("a").values("propertyB").as("propB")).where("propA",
not(P.eq("propB"))).select("a")

>> email to gremlin-user...@googlegroups.com.

>> To view this discussion on the web visit
>> https://groups.google.com/d/msgid/
>> gremlin-users/50896973.2DW1h1rdbW%40localhost.localdomain.
>> For more options, visit https://groups.google.com/d/optout.
>>
>>
>>
>> --
>> You received this message because you are subscribed to the Google Groups
>> "Gremlin-users" group.
>> To unsubscribe from this group and stop receiving emails from it, send an

>> email to gremlin-user...@googlegroups.com.

>> To view this discussion on the web visit
>> https://groups.google.com/d/msgid/
>> gremlin-users/CA%2Bf9seWqOM%3DXCT0nXuZ%3DfTu3neW_3uiZCsEAQmeHLBqfsB4baw
>> %40mail.gmail.com.
>> For more options, visit https://groups.google.com/d/optout.
>>
>>
>>
>>
>> --
>> Lukas Krejci
>>
>> --
>> You received this message because you are subscribed to the Google Groups
>> "Gremlin-users" group.
>> To unsubscribe from this group and stop receiving emails from it, send an

>> email to gremlin-user...@googlegroups.com.

>> To view this discussion on the web visit
>> https://groups.google.com/d/msgid/gremlin-users/3581542.WupXtlzBPQ%40localhost.localdomain.
>> For more options, visit https://groups.google.com/d/optout.
>
>

> --
> You received this message because you are subscribed to the Google Groups
> "Gremlin-users" group.
> To unsubscribe from this group and stop receiving emails from it, send an

> email to gremlin-user...@googlegroups.com.

> To view this discussion on the web visit

> https://groups.google.com/d/msgid/gremlin-users/CA%2Bf9seXz8yNOGJpPaPvLW3WDNKCAQxfxsPVhJUzo%2B_uwpPKG7A%40mail.gmail.com.

> For more options, visit https://groups.google.com/d/optout.



--

Robert Dale

[Marko Rodriguez]

Hello,

Note that I just pushed a fix to tp31/ and up-merged to master/. Thus, your traversal will work as expected in 3.1.5 and 3.2.3.

https://github.com/apache/tinkerpop/commit/c7f42d827414ea7dced0549c02bbc37f40666849

gremlin> g = TinkerFactory.createModern().traversal()

==>graphtraversalsource[tinkergraph[vertices:6 edges:6], standard]

gremlin> g.V().not(match(

......1>     __.as("a").values("age").as("b"),

......2>     __.as("a").values("name").as("c")).

......3>       where("b", eq("c")).

......4>     select("a")).

......5>   values("name")

==>marko

==>vadas

==>lop

==>josh

==>ripple

==>peter

gremlin>

Thanks for the find,

Marko.

http://markorodriguez.com

To unsubscribe from this group and stop receiving emails from it, send an email to gremlin-user...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/gremlin-users/CA%2Bf9seXz8yNOGJpPaPvLW3WDNKCAQxfxsPVhJUzo%2B_uwpPKG7A%40mail.gmail.com.