Extracting data during link walking using PBC

OJ Reeves oj at buffered.io
Sun May 1 19:29:40 EDT 2011

Hi Russell,

Thanks for the response. Pardon my tardiness in responding, I've had lots of
fun at home with sick kids of late.

Responses inline.

On 28 April 2011 16:46, Russell Brown <russell.brown at me.com> wrote:

> Hi OJ!
> I'm just working through this myself at the moment as I'm trying to make
> the Java PB client behave the same way as the REST client .
> On 28 Apr 2011, at 04:23, OJ Reeves wrote:
> So my question: Is this how link walking should work when using the PBC? Or
> is there some way of getting the PBC to return the content at the same time
> as doing the link walking?
> I asked yesterday and the answer I received was: no.

Cool, so I'm not barking mad then! Excellent :) Did you get any feedback as
to whether or not there would be future work done to enable this?

> In practice, what I have it a relationship between three entities (*a*, *b
> * and *c*) that live in different buckets (*A*, *B* and *C*). *a* has a
> link to *b*, and *b* has a link to *c*. What I want to do is walk from *a*to
> *c* while taking the following information:
>    - The value stored in *a*.
>    - The key of *b* and, if *b* still exists in *B*, the value stored in *
>    b*, otherwise nothing/false/failure.
>    - The key of *c* and an indication of whether *c* currently exists in *
>    C*.
> Currently I link walk and do each step manually and I was wondering if
> there was a better or more optimal way of performing these actions.
> At the moment the plan is running an initial m/r job for the link phases
> then making a second m/r job with all the intermediate results as inputs. So
> maximum 2 jobs is my target.
> I should think this is a common problem so if you come up with an optimal
> solution you'll make a few friends!

Off the top of my head with the current implementation I can't see how you
would do this any better. In fact, for link walks where the levels > 2 this
is really the only option (IMHO). For 2 levels or less you can easily fetch
based on ID and you end up with the data faster with the same number of

I will think about it a bit more and see what I can come up with (as soon as
this sickness clears!).

> How do you satisfy your requirement with the REST link walker? Specifically
> how do you get the value of *a* returned as part of the result?

With the REST link walker I'm not able to do this, so I have to make another
call to get to this info. It'd be nice if the link walking mechanism also
had a flag that said "please give me the link source value too". I realise
that might be pie in the sky though :)

Cheers mate, thanks for the response.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.basho.com/pipermail/riak-users_lists.basho.com/attachments/20110502/4f5881dc/attachment.html>

More information about the riak-users mailing list