Sybase offset for pagination

Question

Is there any simple way to implement pagination in sybase? In postgres there are limit and offset in mysql there is limit X,Y. What about sybase? There is top clausure to li

Answer 1

You could try using the set ROWCOUNT twice like this:

    declare @skipRows int, @getRows int
    SELECT @skipRows=50
    SELECT @getRows=10
    set ROWCOUNT @skipRows
    SELECT caslsource_id into #caslsource_paging FROM caslsources
    set rowcount @getRows
    Select * from caslsources where caslsource_id not in (select caslsource_id from #caslsource_paging)
    DROP TABLE #caslsource_paging

This creates a temporary table of rows to skip. You will need to add your WHERE and ORER BY clauses to both the SELECTs to skip the right pages.

Answer 2

I don't know if this is ASE or a different product but the following pattern works across several databases I've worked with as long as there is a way to produce a temp table with a row number somehow and you can identify a unique key for each row:

Input parameters:

declare @p_first int /* max number of rows to see; may be null (= all results); otherwise must be positive number */
declare @p_skipFirst int /* number of rows to skip before the results; must be nonnegative number */
declare @p_after PKTYPE /* key for the row before you start skipping; may be null */

given a table:

RowNumber | RowIndex | DataCol1
1         | 1234     | Joe
2         | 1235     | Sue
3         | 2000     | John
4         | 2005     | Frank
5         | 3000     | Tom
6         | 4000     | Alice

the parameter set:

set @p_first = 5
set @p_skipFirst = 2
set @p_after = 1235

would represent rows 5 and 6.

An additional set of parameters can represent paging from the end of the table in reverse:

declare @p_last int /* max number of rows to see; may be null (= all results); otherwise must be positive number */
declare @p_skipLast int /* number of rows to skip after the results; must be nonnegative number */
declare @p_before PKTYPE /* key for the row after you start skipping; may be null */

Assuming your unsorted table is in #resultsBeforeSort with an index column named RowIndex you can sort this with the following script:

select RowNumber = identity(10), * 
into #results
from #resultsBeforeSort
/*
you might also wish to have a where clause on this query

this sort is dynamically generated based on a sort expression and 
ultimately ended with RowIndex to ensure a deterministic order
*/
order by Column1, Column2 desc, RowIndex

declare @p_total int, @p_min int, @p_max int
select @p_total = count(*) from #results

select @p_min = case when @p_after is null then 1 + @p_skipFirst else @p_total + 1 end
select @p_min = RowNumber + @p_skipFirst from #results where [RowIndex] = @p_after

select @p_max = case when @p_before is null then @p_total - @p_skipLast else 0 end
select @p_max = RowNumber - @p_skipLast from #results where [RowIndex] = @p_before

declare @p_min2 int, @p_max2 int
set @p_min2 = @p_min
set @p_max2 = @p_max

select @p_max2 = case when @p_first is null then @p_max else @p_min + @p_first - 1 end
select @p_min2 = case when @p_last is null then @p_min else @p_max - @p_last end

select @p_min = case when @p_min2 > @p_min then @p_min2 else @p_min end
select @p_max = case when @p_max2 < @p_max then @p_max2 else @p_max end

that script sets up the parameters @p_min, @p_max, and @p_total as well as the temp table #results

You can then use this to select actual data; select 2 table results, first one being metadata (select this first because the second table might not have any actual rows and your reader implementation might not be capable of dealing with that without backtracking):

select [Count] = @p_total, 
    HasPreviousPage = (case when @p_min > 1 then 1 else 0 end),
    HasNextPage = (case when @p_max + 1 < @p_total then 1 else 0 end)

followed by the paged window of results that you actually want:

select [RowIndex], Col1, Col2, Col3 
from #results where RowNumber between @p_min and @p_max

Doing this generic solution permits the ability to expose whatever paging strategy you wish. You can do a streaming solution (facebook, google, stackoverflow, reddit, ...) via @p_after and @p_first (or @p_before and @p_last). You can do an offset + take with @p_first and @p_skipFirst. You can also do a page + size with the same parameters @p_first = size and @p_skipFirst = (page - 1) * size. Further you can do more esoteric paging strategies (last X pages, between absolute records, offset + anchor, etc) with other combinations of parameters.

---

This said, Sybase (SAP) ASE does now directly support the offset + take strategy via rows limit @p_first offset @p_skipFirst. If you only wished to support that strategy you could simplify the above to:

declare @p_total int
select @p_total = count(*) from #resultsBeforeSort

select [Count] = @p_total,
       [HasPreviousPage] = (case when @p_skipFirst > 0 then 1 else 0 end),
       [HasNextPage] = (case when @p_total > @p_skipFirst + @p_first then 1 else 0 end)

select [RowIndex], Col1, Col2, Col3
  from #resultsBeforeSort
  order by Column1, Column2 desc, RowIndex
  rows limit @p_first offset @p_skipFirst

Answer 3

Quoting from http://www.isug.com/Sybase_FAQ/ASE/section6.2.html#6.2.12:

Sybase does not have a direct equivalent to Oracle's rownum but its functionality can be emulated in a lot of cases.

You can set a maximum rowcount, which will limit the number of rows returned by any particular query:

set rowcount 150

That limit will apply until it is reset:

set rowcount 0

You could select into a temporary table, then pull data from that:

set rowcount 150

select pseudo_key = identity(3),
       col1,
       col2
  into #tempA
  from masterTable
 where clause...
 order by 2,3

select col1,col2 from #tempA where pseudo_key between 100 and 150

You could optimize storage on the temp table by storing only ID columns, which you then join back to the original table for your select.

The FAQ also suggests other solutions, including cursors or Sybperl.

Answer 4

Sybase SQL Anywhere example, rows per page:10, offset:1000.

SELECT top 10 start at 1001  * FROM employee order by employeeid

Note: You need to specify the order by column.

Answer 5

// First row = 1000
// Last row = 1009
// Total row = 1009 - 1000 + 1 = 10
// Restriction: exec sp_dboption 'DATABASE_NAME','select into/bulkcopy','true'
select TOP 1009 *, rownum=identity(10) 
into #people
from people 
where upper(surname) like 'B%'
select * from #people where rownum >= 1000
drop table #people
// It shoulde be better SQL-ANSI-2008 (but we have to wait):
// SELECT * FROM people
// where upper(surname) like 'B%'
//    OFFSET 1000 ROWS FETCH NEXT 10 ROWS ONLY

Answer 6

I'm very late to the party but I've happened to stumble on this problem and found a better answer using TOP and START AT from sybase doc. You need to use ORDER BY for or you will have unpredictable results.

http://dcx.sybase.com/1101/en/dbusage_en11/first-order-formatting.html

SELECT TOP 2 START AT 5 * FROM Employees ORDER BY Surname DESC;