[10g] Way to find least value greater than a value?

Question

I have a simple (I think) question ...
Part 1:
Is there a way to determine the least value, among a set of values, that is greater than 0?
A couple of examples:
Set of values:
{0,1,2,3}
Return value:
1
Set of values:
{0,5,10,20,100}
Return value:
5
Set of values:
{1,4,9,11}
Return value:
1Part 2:
Same thing, but greater than some given value, not necessary 0.
Set of values:
{0,1,2,3}
Parameter value:
2
Return value:
3
Set of values:
{0,5,10,20,100}
Parameter value:
2
Return value:
5
Set of values:
{1,4,9,11}
Parameter value:
2
Return value:
4In particular, I'm looking for an efficient way of doing this. I believe I can do it with a set of CASE statements, but it gets really complex the more values you are comparing.
My first thought was to use the LEAST function, but that doesn't allow me a way to use a parameter.
Any suggestions?

Answer 1

-- this section has been edited
And that's basically where I was heading. I was hoping there was a less complex way to do it... somehow replacing any zeros with an impossibly large number, without using case statements (something equivalent to the NVL function, but to replace any value with another), so that they'd never be less than anything, and then I could just use the LEAST function. The only way I can think of to do that though, would be to convert the numbers to strings, use REPLACE, and then convert them back to numbers, which doesn't seem less complex than the CASE statements, and most likely is worse performance-wise.
For example:
SELECT     ord_nbr
,     seq_nbr
,     time_spent /
     CASE     WHEN     comp_qty     <= plan_qty
          AND     comp_qty     <> 0
          THEN     comp_qty
          WHEN     comp_qty     > plan_qty
          THEN     LEAST     ( TO_NUMBER(REPLACE(TO_CHAR(ord_qty-scrap_qty),'0','999999999'))
                    , TO_NUMBER(REPLACE(TO_CHAR(plan_qty),'0','999999999'))
                    , TO_NUMBER(REPLACE(TO_CHAR(comp_qty),'0','999999999'))
          ELSE 1
     END     AS unit_time
FROM     ord_detail
;-- end of section edited (I posted before I had fully gone through the solution provided)
I checked the data, and though, theoretically all values could be zero, there are no instances where that happens. I think that, in the case all values were 0, I'd just want to use 1 instead.
Side note on the "big" problem behind this question....
My ultimate problem, (and if I end up needing help with that, I'll start a new post for it) is that the quantity data in our system has a fair amount of junk in it...values that can't be true. With this solution, I would assume that in any case where the quantity complete at a given step of a process is less than or equal to the quantity planned to be complete at that step, the value is good, which is not necessarily a correct assumption. Then, only in cases where something can't be true, like when quantity complete > the quantity planned, are when I would intervene with this solution to make a "best guess" at the correct value. There a few things related to this that I have to determine before I can figure out my next step.
We have another database (so we can keep things straight, I'll call the database I've been querying up til now DB1, and this other one, DB2) that has much more reliable quantity data and step data (sometimes it has steps that aren't in DB1) but its structure is complex, I'm rather unfamiliar with it, and it does not have time spent data. Additionally, some products have all their information in DB1, and none in DB2.
So my options are to:
1) ...try to learn the structure of the other database, find the data I need, and query both databases together, using this thread's solution to resolve any questionable data that does not exist in both systems, and skipping any steps that occur in DB2, but not DB1 (since they won't have any time data, which is ultimately what I'm after)
2) ...try to come up with a method to pre-"scrub" my data in DB1, based on some logical assumptions and looking at all steps in an order, so that I can then just query the scrubbed data...
3) ...use the solution from this thread, and assume that the bad data isn't enough to significantly impact my calculations...
What I wouldn't give for a single system with all good data!
Edited by: user11033437 on Nov 19, 2010 4:21 PM