What's the most efficient way of starting a Gremlin query from a list?

[James Thornton]

You can start with a range of vertices like this:

gremlin> g.V[1..2]

==>v[2]

==>v[1]

But how do you start it with a list?

gremlin> g.V[1,2,3,5,8,13,21,34,55,89,144] 

No signature of method: com.tinkerpop.gremlin.pipes.GremlinPipeline.getAt() is applicable for argument types: (java.util.ArrayList) values: [[1, 2, 3, 5, 8, 13, 21, 34, 55, 89, 144]]

Possible solutions: getAt(java.lang.String), getAt(groovy.lang.Range), getAt(java.util.Map), getAt(java.lang.String), getAt(int), getAt(java.lang.Integer)

Display stack trace? [yN] 

[Pierre De Wilde]

Hey,

In your examples, it is not clear if you specify the index value or the vertex id. 

I don't see an use case with the index value since g.V returns all vertices in an undefined order.

Anyway, to start a Gremlin statement with a list of vertices using index value:

i=0

g.V.filter{i++ in [1,2,3,5,8,13,21,34,55,89,144]}

To start a Gremlin statement with a list of vertices using vertex id:

g.V.filter{it.id in [1,2,3,5,8,13,21,34,55,89,144]}

A far more efficient version using vertex id is

[ g.[1],g.v[2],g.v[3],g.v[5],g.v[8],g.v[13],g.v[21],g.v[34],g.v[55],g.v[89],g.v[144] ]._()

because instead of sequential filtering, you only get what you want.

HTH,

Pierre

[James Thornton]

Hi Pierre -

Thanks for help. A use case would be a search engine like Solr returning a list of 1000 IDs that you want to rank and sort.

- James

[Pierre De Wilde]

> A use case would be a search engine like Solr returning a list of 1000 IDs that you want to rank and sort.

Why do you want to apply Fibonacci indices at that ?

[James Thornton]

> A use case would be a search engine like Solr returning a list of 1000 IDs that you want to rank and sort.

I'm thinking of ways to apply Marko's ideas on "local rank" to search results (http://markorodriguez.com/2011/03/30/global-vs-local-graph-ranking/), where local rank is a subset of all the vertices. Global rank is all vertices, and ego rank is one.

- James

[James Thornton]

> Why do you want to apply Fibonacci indices at that ?

No particular reason -- I was just using it as an example of a non-sequential list :)

[Marko Rodriguez]

Hi,

As Pierre says, you can do:

[g.v(1), g.v(2), g.v(10)]._().out.in....

Perhaps a bit verbose? I can add this method to Gremlin:

g.v(1,2,10)._().out.in...

Would that be cool?

I could also turn it into a Pipe and thus:

g.v(1,2,10).out.in...

...however, problem with the last is that then g.v() would have mixed mappings so thats probably not a good idea.

Thoughts?,

Marko.

http://markorodriguez.com

[James Thornton]

> I can add this method to Gremlin:

Hey Marko -

Yeah, that would be totally cool -- whatever you think would be the most performant for a large number of vertices. I'm still exploring this idea, but I think this might be a good way to build ranking algorithms into search.  

- James

[Marko Rodriguez]

Hi,

See https://github.com/tinkerpop/gremlin/wiki/Gremlin-Methods

You can now do:

gremlin> g.v(1) ==>v[1] gremlin> g.v(1,2,3) ==>v[1] ==>v[2] ==>v[3] gremlin> g.v(1,2,1000) ==>v[1] ==>v[2] ==>null gremlin> g.e(7,8) ==>e[7][1-knows->2] ==>e[8][1-knows->4]

If you pass in one id (e.g. g.v(1)), you will get back a single element. If you pass in more than one id (e.g. g.v(1,2)), you will get back a list of elements.

This has been pushed to GitHub in Gremlin 1.3-SNAPSHOT.

Enjoy,

Marko.

http://markorodriguez.com

[James Thornton]

> If you pass in one id (e.g. g.v(1)), you will get back a single element. 

> If you pass in more than one id (e.g. g.v(1,2)), you will get back a list of elements.

> This has been pushed to GitHub in Gremlin 1.3-SNAPSHOT.

That was fast! :) Thanks man.

- James

[Pierre De Wilde]

Hey,

Cool.

Why g.v(id,id) does return LinkedList instead of FluentPipeline ?

g.v(1).out but g.v(1,2)._().out but g.idx(T.v)[[name:'josh']].out

Thanks,

Pierre

[Marko Rodriguez]

Hi,

gremlin> g.v(1).class

==>class com.tinkerpop.blueprints.pgm.impls.tg.TinkerVertex

gremlin> g.v(1,2).class

==>class java.util.LinkedList

gremlin> g.idx(T.v)[[name:'josh']].class

==>class com.tinkerpop.pipes.util.FluentPipeline

g.v() are  not pipelines. I can make g.v(1,2) a pipeline... ? .. .? . ... The reason I didn't was for stuff like this:

g.v(1).out.out.except(g.v(2,3,4)).in

In other words, g.v() is just a "getter" of vertices, not a pipeline construct.

Marko.

http://markorodriguez.com

[Pierre De Wilde]

Hi Marko,

Both makes sense but I concur by your last argument: g.v() is just a "getter" of vertices.

g.v(2,3,4)._() implicit notation is definitively not my cup of (Oolong) tea.

Therefore, I reiterate an old request to use a more explicit .toPipe() instead of / in complement to  ._() notation.

Thanks,

Pierre

[Peter Neubauer]

+1

On toPipe
Also, I would love to make >> explicit like emit() or so. These two constructs lead to a lot of errors and confusion...

/Peter

On Sunday, August 28, 2011, Pierre De Wilde <pierre...@gmail.com> wrote:
> Hi Marko,
> Both makes sense but I concur by your last argument: g.v() is just a "getter" of vertices.

> g.v(2,3,4)._() implicit notation is definitively not my cup of (Oolong <http://en.wikipedia.org/wiki/Oolong>) tea.
> Therefore, I reiterate an old request <https://github.com/tinkerpop/gremlin/issues/203> to use a more explicit .toPipe() instead of / in complement to  ._() notation.

> Thanks,
> Pierre
>
>
> On Sun, Aug 28, 2011 at 2:23 PM, Marko Rodriguez <okram...@gmail.com> wrote:
>>
>> Hi,
>> gremlin> g.v(1).class
>> ==>class com.tinkerpop.blueprints.pgm.impls.tg.TinkerVertex
>> gremlin> g.v(1,2).class
>> ==>class java.util.LinkedList
>> gremlin> g.idx(T.v)[[name:'josh']].class
>> ==>class com.tinkerpop.pipes.util.FluentPipeline
>> g.v() are  not pipelines. I can make g.v(1,2) a pipeline... ? .. .? . ... The reason I didn't was for stuff like this:
>> g.v(1).out.out.except(g.v(2,3,4)).in
>> In other words, g.v() is just a "getter" of vertices, not a pipeline construct.
>> Marko.

>> http://markorodriguez.com <http://markorodriguez.com>

>> On Aug 28, 2011, at 1:33 AM, Pierre De Wilde wrote:
>>
>> Hey,
>> Cool.
>> Why g.v(id,id) does return LinkedList instead of FluentPipeline ?
>> g.v(1).out but g.v(1,2)._().out but g.idx(T.v)[[name:'josh']].out
>> Thanks,
>> Pierre
>>
>> On Sat, Aug 27, 2011 at 5:59 PM, James Thornton <james.t...@gmail.com> wrote:
>>>
>>> > If you pass in one id (e.g. g.v(1)), you will get back a single element. 
>>> > If you pass in more than one id (e.g. g.v(1,2)), you will get back a list of elements.
>>> > This has been pushed to GitHub in Gremlin 1.3-SNAPSHOT.
>>> That was fast! :) Thanks man.
>>> - James
>>
>
>

--

Cheers,

/peter neubauer

GTalk:      neubauer.peter
Skype       peter.neubauer
Phone       +46 704 106975
LinkedIn   http://www.linkedin.com/in/neubauer
Twitter      http://twitter.com/peterneubauer

http://www.neo4j.org               - Your high performance graph database.
http://startupbootcamp.org/    - Öresund - Innovation happens HERE.
http://www.thoughtmade.com - Scandinavia's coolest Bring-a-Thing party.

[James Thornton]

> I would love to make >> explicit like emit() or so. These two constructs lead to a lot of errors and confusion...

I agree, the >> too often leads to confusion. You can use toList() in its place sometimes, but a more explicit alternative for >> would help.

- James

[Marko Rodriguez]

Hi,

I am not a fan of toPipe(). I could do something like:

start(g.v(2,3,4)).out.in.out...

Which would be a static method reference to:

new FluentPipeline().start(g.v(2,3,4)).out.in.out...

Marko.

http://markorodriguez.com

[Marko Rodriguez]

Hi Peter,

You can do:

g.v(1).out.out.iterate()  <=> g.v(1).out.out >> -1

g.v(1).out.out.toList() <=> g.v(1).out.out >> []

g.v(1).out.out.next() <=> g.v(1).out.out >> 1

g.v(1).out.out.next(5) <=> g.v(1).out.out >> 5

*** NOTE: next(int number) was recently implemented, but not pushed GitHub or Maven2.

Hope that is sufficient for your cares,

Marko.

http://markorodriguez.com

[Peter Neubauer]

Yup,
Looks good!

/peter

>>> http://markorodriguez.com <http://markorodriguez.com/> <http://markorodriguez.com <http://markorodriguez.com/>>

>>> On Aug 28, 2011, at 1:33 AM, Pierre De Wilde wrote:
>>>
>>> Hey,
>>> Cool.
>>> Why g.v(id,id) does return LinkedList instead of FluentPipeline ?
>>> g.v(1).out but g.v(1,2)._().out but g.idx(T.v)[[name:'josh']].out
>>> Thanks,
>>> Pierre
>>>
>>> On Sat, Aug 27, 2011 at 5:59 PM, James Thornton <james.t...@gmail.com> wrote:
>>>>
>>>> > If you pass in one id (e.g. g.v(1)), you will get back a single element. 
>>>> > If you pass in more than one id (e.g. g.v(1,2)), you will get back a list of elements.
>>>> > This has been pushed to GitHub in Gremlin 1.3-SNAPSHOT.
>>>> That was fast! :) Thanks man.
>>>> - James
>>>
>>
>>
>
> --
>
> Cheers,
>
> /peter neubauer
>
> GTalk:      neubauer.peter
> Skype       peter.neubauer

> Phone       +46 704 106975 <tel:%2B46%20704%20106975>

> LinkedIn   http://www.linkedin.com/in/neubauer
> Twitter      http://twitter.com/peterneubauer
>

> http://www.neo4j.org <http://www.neo4j.org/>               - Your high performance graph database.

> http://startupbootcamp.org/    - Öresund - Innovation happens HERE.

> http://www.thoughtmade.com <http://www.thoughtmade.com/> - Scandinavia's coolest Bring-a-Thing party.

[Pierre De Wilde]

Hey,

Thanks for those good alternatives to >> x.

Maybe .next(-1) as an alternative to .iterate() ?

Regarding _() issues, devil advocate put the shoes of newbie and try:

gremlin> g.v(1).out

==>v[2]

==>v[3]

==>v[4]

gremlin> g.v(1,2).out

==>[StartPipe, OutPipe]

==>[StartPipe, OutPipe]

gremlin> g.V.filter{it.id in [1,2]}.out   // no results because ids are strings in TinkerGraph example

gremlin> g.V.filter{it.id in ['1','2']}.out

==>v[2]

==>v[3]

==>v[4]

Newbie is confused, as devil advocate. But not for the same reasons: 

- newbie uses try-and-error to figure out how Gremlin works 

and is confused because he doesn't understand g.v(1,2).out results.

- devil advocate knows that g.v(1,2) returns LinkedList that needs to be converted 

into FluentPipeline using the ._() notation and is confused by this weird syntax.

I definitely need a rest.

Pierre