Re: Would it be a reasonable to ask IBM's programmers to write a%sortsfl(sfl: column)

Trevor, the very vast majority of the subfiles I have ever seen are under 1000 records, most of those are under 500. Yes, I have seen these page-at-a-time subfiles written for files that will never have more than 50 entries. Even large files that are filtered are generally filtered to sizes well below 500 records.

However, your point is well made. That is why I mentioned the %bif being for load-all subfiles only

.

Trevor Perry wrote:

Booth,

For the smarter programmers, the subfile is built one or three pages at a
time. In that case, the program does not have a copy of the entire 9,999
records. Which makes it difficult to sort when you only have a subset of the
entire thing.

Trevor


On 12/8/09 4:44 PM, "Booth Martin" <booth@xxxxxxxxxxxx> wrote:

Thats the part I did not understand.

The subfile is all ready written. It is a known block of data. There's
no need for any under the covers work or any connection to the data;
just process the existing data in the existing subfile. Sort the first
request ascending and any following requests in the opposite order.

Frankly it surprised me that someone didn't say this is a lunch time
sport for the people that are really good at %bifs.




DeLong, Eric wrote:

Hmm, maybe just allow an SQL statement to be associated with the SFL,
and allow OS400 to manage the paging through the result set?

-Eric DeLong