RE: User Index APIs - where find REAL reference material?

>>> -----Original Message-----
>>> From: Dan Bale [mailto:dbale@genfast.com]
>>> Sent: Wednesday, July 28, 1999 10:37 PM
>>> To: MIDRANGE-L@midrange.com
>>> Subject: User Index APIs - where find REAL reference material?
>>> 
>>> 
>>> 
>>> 
>>> I am trying to understand and create my own applications 
>>> using user indexes, but
>>> am relegated to using others' examples, which does not help 
>>> me understand the
>>> intricacies of the APIs involved.
>>> 
>>> For example, on the Create User Index (QUSCRTUI) API, I am 
>>> finding minimal
>>> information on the "Key insertion" and the "Immediate 
>>> Update" parameters.
>>> ---------------------------------------------------------
>>> Key Insertion: Whether or not the inserts to the index are 
>>> by key.  The valid
>>> values are:
>>>      '0'  No insertion by key
>>>      '1'  Insertion by key
>>> 
>>> Is this asking whether I'll be loading the index in key 
>>> sequence?  Am I
>>> confusing the term "insert" with "add"?  (as in "inserting" 
>>> a new entry in the
>>> middle of the index vs. "adding" a new entry at the end of 
>>> the index)

 Yes, I think so, worked for me anyway using '1'

>>> ---------------------------------------------------------
>>> Immediate Update:  Whether or not the updates to the index 
>>> are written
>>> synchronously to auxiliary storage on each update to the 
>>> index.  The valid
>>> values are:
>>>      0    No immediate update
>>>      1    Immediate update
>>>  Each update to the index is written to auxiliary storage 
>>> after every insert and
>>> remove operation.
>>> 
>>> What impact does this parameter have?  In the one 
>>> application, the UI would be
>>> created and loaded with data.  It would not be updated, 
>>> i.e., existing entries
>>> modified or entries added or deleted.

This probably depends on the usage of the data your storing in there.
I used the user index as an alternative to arrays
to check for duplicate entries in subfile validation, so if the
job/system crashed I wouldnt be too bothered if I lost the data,
but if your data was part of a transaction that might be output to a
database at some point, you probably wouldnt want to loose the data if
the
job/system crashed, so would want to update to aux storage
immediately. I presume this is some sort of performance trade-off here.
>>> 

>>> Also, on the Add User Index Entries (QUSADDUI) API, it 
>>> appears that you can add
>>> up to 4095 index entries on one call.  Is this correct?  If 
>>> so, would you
>>> typically use an array or a multiple-occurence data 
>>> structure to pass the data?

I would probably just use a field defined over an array.  

>>> Those of you who have the answers, where did you learn 
>>> this?  I certainly have
>>> failed to find out in the IBM softcopy manuals.

I just used the information in the API manual, the rest is just a
question of trial and error. I got my user index working quite quickly,
and it did
exactly what I needed.

Cheers
Colin.W 
+---
| This is the Midrange System Mailing List!
| To submit a new message, send your mail to MIDRANGE-L@midrange.com.
| To subscribe to this list send email to MIDRANGE-L-SUB@midrange.com.
| To unsubscribe from this list send email to MIDRANGE-L-UNSUB@midrange.com.
| Questions should be directed to the list owner/operator: david@midrange.com
+---