[Spread-users] Order of members in vs_sets

[Yair Amir]

Hi,

The concept of "the oldest member of the group" is not
well defined in a partitionable system.

Cheers,

	:) Yair.

On 3/30/09 5:31 PM, Luke Marsden wrote:
> Hi guys,
> 
> Is there any way to get the order in which nodes joined a group out of
> Spread without "watching" (not always possible)?
> 
> I'd like to use the oldest member of the group as the leader rather than
> the lowest IP address, which is what we're presently using.
> 
> Thanks!
> Luke Marsden
> Hybrid Logic Ltd.
> 
> 
> On Mon, 2009-03-30 at 18:19 -0400, John Schultz wrote:
>> I believe each vs_set is sorted using strcmp as the comparison.  So,
>> within a single vs_set I *believe* they will be sorted in ascending
>> order according to strcmp.
>>
>> I don't know if that means you can use std::set_difference() or not
>> as I'm not intimately familiar with all the STL functions.
>>
>> Cheers!
>> John
>>
>> ---
>> John Lane Schultz
>> Spread Concepts LLC
>> Phn: 443 838 2200 
>> Fax: 301 560 8875
>>
>> Monday, March 30, 2009, 2:07:51 PM, you wrote:
>>
>>> Dear Spread users,
>>> I wonder if the list of members in a vs_set, as returned by 
>>> SP_get_vs_set_members(), is ordered in a deterministic way. The 
>>> documentation for SP_receive() states that the list of members returned
>>> through the groups parameter (the current membership view) is the same
>>> for all the processes receiving the membership message, so that a leader
>>> can be deterministically elected for the group. Does the list of members
>>> is a vs_set follow the same ordering?
>>> In other words, can I use std::set_difference(), between 
>>> SP_get_vs_set_members(vs_sets[my_vs_set]) and the previous membership 
>>> view, to determine the members who are no longer part of the group after
>>> receiving a CAUSED_BY_NETWORK membership message?
>>> Thanks,
>>
>>> Tudor
>>
>>
>> _______________________________________________
>> Spread-users mailing list
>> Spread-users at lists.spread.org
>> http://lists.spread.org/mailman/listinfo/spread-users