order of siblings

Erik Søe Sørensen ess at trifork.com
Sun Mar 25 16:15:52 EDT 2012

I'm not sure what you mean by "assuming the vclock entries are 1-1 with the siblings".
That sounds rather like a misunderstanding of either what the vclock is, or of how the vclock is stored in the case of siblings.
(As for the latter, only the merged vclock is stored for the entire key/value-pair; the length of that vclock is unrelated to the number of siblings.)

Mobile: + 45 2636 1755 | Skype: eriksoesorensen | Twitter: @eriksoe
Trifork A/S  |  Margrethepladsen 4  | DK- 8000 Aarhus C |  Phone : +45 8732 8787  |  www.trifork.com
From: riak-users-bounces at lists.basho.com [riak-users-bounces at lists.basho.com] On Behalf Of Michael Radford [mrad at blorf.com]
Sent: 24 March 2012 19:24
To: riak-users at lists.basho.com
Subject: order of siblings

Is there a way to determine what Riak thinks is the last-written
sibling (or one of them), when reading from Riak with allow_mult=true
(via the erlang protobufs api)?

I'm writing some conflict resolution code, which in many cases should
fall back to the default strategy of last-writer-wins, and it would be
nice to use Riak's default ordering instead of embedding an extra
timestamp in the value.

It looks like something like the unpacking of the vclock here might work:

...assuming the vclock entries are 1-1 with the siblings, but that's
not clear to me. Or maybe the siblings already come in a particular
order, oldest or newest first?


riak-users mailing list
riak-users at lists.basho.com

More information about the riak-users mailing list