Actually what I've done is construct the adjacency list prior to running the job. The data sets so far are not very big so a simple Perl script did the trick. For larger data sets I'll use MapReduce.
Thank you, Steve From: Avery Ching [mailto:ach...@apache.org] Sent: Friday, March 16, 2012 2:06 AM To: giraph-user@incubator.apache.org Subject: Re: Questions about the shortest path example If I'm understanding your question correctly, when you load the vertex, you'll want to read all the lines of the same vertex id and then construct the vertex with the edges. The edges of the vertex need not be sorted. But the lines of the same vertex id will have to be grouped together so that you can process one vertex at a time. This is the case until something like GIRAPH-155 is completed. Avery On 3/15/12 4:14 PM, Fleischman, Stephen (ISS SCI - Plano TX) wrote: I'm working with the shortest path example and using datasets consisting of two columns - a "from" vertex and a "to" vertex, so that a vertex with n edges is repeated n times. In constructing the sorted adjacency list does each adjacent vertex array also have to be sorted? Thank you, Steve Fleischman