hama-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Edward J. Yoon" <edwardy...@apache.org>
Subject Re: Contributing to project Hama as a group
Date Tue, 26 Aug 2014 22:43:45 GMT

HAMA-891 might be tricky to implement, but will be really meaningful!

The greedy heuristic partitioning is done by assigning the vertex to
the partition where it has the most edges. Please find the "(Weighted)
Deterministic Greedy" part in

The tricky issue is that you'll need to create something for managing
the globally-accessable table of vertex IDs to partition IDs. Because,
current system always calls the getPartitionID() method when sending a
message (this is make sense only when the calculation is simple like

  public void sendMessage(Edge<V, E> e, M msg) throws IOException {
        new GraphJobMessage(e.getDestinationVertexID(), msg));

   * @return the destination peer name of the given vertex id, determined by the
   *         partitioner.
  public String getDestinationPeerName(V vertexId) {
    return runner.getPeer().getPeerName(
        getPartitioner().getPartition(vertexId, value,

On Wed, Aug 27, 2014 at 5:22 AM, Bhathiya Supun <hsbathiya@gmail.com> wrote:
> We, group of students from University of Moratuwa are interested
> in contributing project Hama as a part of our Intelligent System module.
> Going through the project issues we found
> https://issues.apache.org/jira/browse/HAMA-891 would be a good fit.
> We would like to discuss possibility of working on this issue (or any other
> issue that would fit). Any advice on getting started is welcome as non of
> us are much familiar with the project.Further, we like to know whether the
> irc channel #hama on freenode is still active.
> Thank you
> Bhathiya

Best Regards, Edward J. Yoon
CEO at DataSayer Co., Ltd.

View raw message