Using &NDBSCAN Statements › Scan Processing › Process Scan Results › Differences Between &NDBSCAN Sequences and &NDBSEQ Sequences

Differences Between &NDBSCAN Sequences and &NDBSEQ Sequences

There are some important differences between sequences defined by &NDBSEQ, and sequences built by &NDBSCAN:

• &NDBSEQ-defined sequences do not have an in-storage list of records associated with them. Thus, skipping forward and backward automatically takes into account record deletions. For example, if you are reading by RID, and you are positioned on RID 10, a GET FWD SKIP=5 returns RID 15, assuming RIDS 11-14 all exist. If, while positioned on RID 15, RIDs 12 and 14 are deleted, a GET BWD SKIP=5 does not position you on RID 10, but, rather, RID 8 (assuming 8 and 9 exist).

  &NDBSCAN-built sequences are represented as an in-storage list of RIDs (the actual order depending on SORT, and so on). If the list is ordered on RID, the previous example repositions from 10 to 15, and back to 10, as SKIP=n instructs &NDBGET to skip over n-1 entries in the list. Only when the target RID has been deleted, does &NDBGET proceed to the next RID in the list, until a non-deleted record is found.

• &NDBSEQ-defined sequences each take a small, fixed amount of storage to hold information about the sequence.

  &NDBSCAN-built sequences take storage proportional to the number of matching records to hold the list. Thus, you should try to minimize the number of concurrent scan sequences in use, particularly if using &NDBSCAN in an EASINET procedure.