Hi, Petr,

Looks like you are using FastBit correctly, we must have something
wrong in handling the column names as you suspected.

Sounds like you have a test program to reproduce this issue.  If you
can share the test program, it would make the debugging work a little
easier for us.

Thanks.

John


On 9/18/14 1:23 AM, Petr Velan wrote:
> Hi John,
> 
> we run into select columns ordering issue while using fastbit. We are
> currently running revision 745. Here is the description of the problem:
> 
> Using ibis tool in interactive mode:
> ibis -i
> load the part
> 
>     append 257
>     print parts
> 
> Name(s) of all data partitioins
> M35323568
> 
> Do select, which fails:
> 
>     SELECT e0id152, (e0id153 - e0id152) as td, e0id4, e0id8, e0id7,
>     e0id12, e0id11, e0id7, e0id11, e0id6, e0id5, e0id2, (e0id1 /
>     (e0id153 - e0id152)) as bps, e0id1 FROM M35323568
> 
> tableSelect:: select(e0id152, (e0id153 - e0id152) as td, e0id4, e0id8,
> e0id7, e0id12, e0id11, e0id7, e0id11, e0id6, e0id5, e0id2, (e0id1 /
> (e0id153 - e0id152)) as bps, e0id1, 1=1) failed on table T-M35323568
> 
> Just by moving second to last column to front, the query works:
> 
>     SELECT (e0id1 / (e0id153 - e0id152)) as bps, e0id152, (e0id153 -
>     e0id152) as td, e0id4, e0id8, e0id7, e0id12, e0id11, e0id7,
>     e0id11, e0id6, e0id5, e0id2, e0id1 FROM M35323568
> 
> tableSelect -- select((e0id1 / (e0id153 - e0id152)) as bps, e0id152,
> (e0id153 - e0id152) as td, e0id4, e0id8, e0id7, e0id12, e0id11, e0id7,
> e0id11, e0id6, e0id5, e0id2, e0id1, 1=1) on table T-M35323568 produced
> a table with 76 rows and 12 columns
> tableSelect -- the first row (of 76) from the result table for "SELECT
> (e0id1 / (e0id153 - e0id152)) as bps, e0id152, (e0id153 - e0id152) as
> td, e0id4, e0id8, e0id7, e0id12, e0id11, e0id7, e0id11, e0id6, e0id5,
> e0id2, e0id1 FROM T-M35323568 WHERE 1=1"
> 3.34857142857143, 1408440725886, 175, 6, 2481262304, 40733,
> 2481259049, 5222, 40733, 5222, 24, 0
> 
> Removing some of the columns without reordering the positions, the
> query works:
> 
>     SELECT e0id152, (e0id153 - e0id152) as td, (e0id1 / (e0id153 -
>     e0id152)) as bps, e0id1 FROM M35323568
> 
> tableSelect -- select(e0id152, (e0id153 - e0id152) as td, (e0id1 /
> (e0id153 - e0id152)) as bps, e0id1, 1=1) on table T-M35323568 produced
> a table with 76 rows and 4 columns
> tableSelect -- the first row (of 76) from the result table for "SELECT
> e0id152, (e0id153 - e0id152) as td, (e0id1 / (e0id153 - e0id152)) as
> bps, e0id1 FROM T-M35323568 WHERE 1=1"
> 1408440725886, 175, 3.34857142857143, 586
> 
> Are we misusing it somehow, or is it a bug?
> 
> Thanks,
> Petr
> 
> 
> _______________________________________________
> 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