How do I efficiently get the first n rows of a query from partitioned table?

[Pavel Hejbal]

I want to get for example first 10 rows from partitioned table in specific date (date is a partition column). The solution is

10# select from trade where date = 2014.01.02

but this query has the same usage of memory as

select from trade where date = 2014.01.02

(I used \ts and get same values for both cases)

so I hope, theres more efficient solution. Any ideas, please?

Thanks Pavel

[Charles Skelton]

This isn't necessary on 64bit, but on 32bit this should reduce your mem load; using functional select, load a column at a time. Here using trade table, last date. Use parse"select …." to figure out how to form functional select.

q)flip raze {flip 10#?[trade;enlist(=;`date;2014.01.02);0b;(enlist x)!enlist x]}each cols trade

--
You received this message because you are subscribed to the Google Groups "Kdb+ Personal Developers" group.
To unsubscribe from this group and stop receiving emails from it, send an email to personal-kdbpl...@googlegroups.com.
To post to this group, send email to personal...@googlegroups.com.
Visit this group at http://groups.google.com/group/personal-kdbplus.
For more options, visit https://groups.google.com/d/optout.

[Sean O'Hagan]

select from trade where date=2014.01.02,i within (1;1+10)

or look at 

http://code.kx.com/wiki/DotQ/DotQDotind

Thanks,

Sean

[Attila Vrabecz]

address-space/mapping is the problem

hence Charlie's suggestion

10#..

or

..i<10

does not matter

Cheers,

  Attila

  Attila

--

[Pavel Hejbal]

I can't use i<10 nor select [10] becase it is partitioned table. It tried .Q.ind before, but I waited for resuresult almost half an hour.Selecting by date is quite complicated with .Q.ind and not efficient (searching for starting index of the particilar date).

[Charles Skelton]

With 32bit, for ease of querying, you probably don't want to go above ~500MB per partition.

See http://code.kx.com/wiki/JB:KdbplusForMortals/segments#1.4_Segments

On Fri, May 23, 2014 at 7:11 PM, Pavel Hejbal <pavel....@gmail.com> wrote:

I can't use i<10 nor select [10] becase it is partitioned table. It tried .Q.ind before, but I waited for resuresult almost half an hour.Selecting by date is quite complicated with .Q.ind and not efficient (searching for starting index of the particilar date).

[Pavel Hejbal]

Thank you for all responses. So it looks, there is no less memory consuming solution, the only way is to keep partiones "small" enough.

Dne pátek, 23. května 2014 19:29:22 UTC+2 Charles Skelton napsal(a):