Thanks John,

I've implemented row sampling after calling query::getQualifiedLongs()
instead of using RIDs.  However, I noticed that
ibis::part::getRIDs().size() == 0, and that in ibis::part() there is a
check for readonly == false and a config key called <table name>.fillRIDs
before the rids are filled.  I figured one of these conditions were not
met.  But are you suggesting to simply call ibis::query::setRIDs() on a
vector of RIDs of my own choosing, without calling ibis::part::getRIDs(),
and assuming they are numbered 0..nRows()-1?

On 8/28/13 10:53 AM, "K. John Wu" <[email protected]> wrote:

>Hi, Steven,
>
>The function ibis::part::hasRIDs is badly named.  It is meant to
>determine if there is an explicit RID column for the data partition.
>It is absolutely fine for a data partition to not have an explicit RID
>column, the implicit values of [0, 1, 2, ...] will be used.  This is
>what happens when you ask for RIDs through getRIDs.  The same
>assumption is also used in answering queries.  So if you simply ignore
>what ibis::part::hadRIDs return to you and go ahead ask the query
>object to evaluate itself, you should get the right answers.
>
>Let us know if the answers are not what you've expected.
>
>Good luck.
>
>John
>
>PS: I am changing hasRIDs to explicitRIDs.  It appears to be one of
>those functions not actually used in FastBit itself..
>
>
>On 8/21/13 2:26 PM, Enns, Steven wrote:
>> Hi John,
>> 
>> I'm looking to sample rows in ibis::part.  The best way that I can
>> think to do so is to call ibis::part::getRIDs(), sample the RIDs, then
>> call ibis::query::setRIDs() to filter the rows.
>> 
>> The problem is that ibis::part::hasRIDs() is false, despite readOnly
>> being false and <tableName>.fillRIDs=T in my ~/.ibisrc file.
>> 
>> Is there some way to write the rids when creating the table?
>> 
>> Thanks,
>> Steve
>> 
>> 
>> _______________________________________________
>> FastBit-users mailing list
>> [email protected]
>> https://hpcrdm.lbl.gov/cgi-bin/mailman/listinfo/fastbit-users
>> 

_______________________________________________
FastBit-users mailing list
[email protected]
https://hpcrdm.lbl.gov/cgi-bin/mailman/listinfo/fastbit-users

Reply via email to