Re: [chiPHPug-discuss] a bit of a sticky wicket...

[Peter H. <ph...@ma...>]

On Fri, Oct 12, 2007 at 12:28:45AM -0500, jsWalter wrote:
> Here is a sample record set...

Would an SQL query to do this be useful, or do you only have this data
in an array?

> 	ID         DATE     line_num  seq_num  cell_num  time
> 1	APM	2007-10-01     1         0         0     8.00
> 2	APM	2007-10-01     1         1         0     8.00
> 3	APM	2007-10-02     1         0         0     0.50
> 4	APM	2007-10-02     2         0         0     6.50
> 5	APM	2007-10-02     2         0         1     6.00
> 6	APM	2007-10-02     2         0         2     5.50
> 7	APM	2007-10-02     3         0         0     1.00
> 8	APM	2007-10-02     3         0         1     0.50
> 9	APM	2007-10-02     4         0         0     1.00
> 10	APM	2007-10-02     4         0         1     1.25
> 11	APM	2007-10-03     1         0         0     1.25
> 12	APM	2007-10-03     1         0         0     2.25
> 
> What I need to do is pull out it the time for each day. This example has 3
> days.
> 
> If you look at line 1 and 2, I would use line #2 because it has 2 LINE_NUM
> that are the same, but the second SEQ_NUM is higher. So I keep it.
> 
> Lines 3 through 12 are the same date, with 5 sets of LINE_NUMs.
>  - #3 we keep, it is unique
>  - #6 we keep, because it's CEL_NUM is highest, even though it has the
>    same LINE_NUM and SEQ_NUMs
>  - #8 & #10 we keep, same reason as #6
>  - we keep both 12 & 12 because the have the same numbers all across the 3
> fields.

So you're selecting the rows for each date and line_num that have (or
tie for) the highest line_num and seq_num. 

You don't mention whether cel_num or seq_num has higher priority.
Which of these two rows would be selected?

 	ID         DATE     line_num  seq_num  cell_num  time
1	APM	2007-10-02     2         0         1     6.00
2	APM	2007-10-02     2         1         0     5.50

--
Peter Harkins  -  http://push.cx  -  http://NearbyGamers.com