WRKOBJLCK automated API ?

[MacWheel99@xxxxxxx]

From Al Macintyre, still writing in RPG/400 but occasionally embedded SQL

This is not urgent, more like brain storming.  I have some software to 
modify, to enhance what it is doing in the form of features, new data 
capturing, less hassles for users & so forth, and one worrysome situation is 
risk of conflict updating a heavily used file.  I would like my next 
modifications to handle that situation more gracefully.

What happens today is that there is lock up because two people are trying to 
update the same file at the same time, and the end user messages are not 
friendly informative.  At best we are told that some other unspecified user 
is updating a record in a specific file that this one needs, & usually does 
not give a clue as to which one, & when it does it gives record # of the 
file, when our users think of the data by item # order # etc., and then 
manual usage of WRKOBJLCK to locate who that is can be a royal pain.  

Finally we figure out who is the blocking program, and get that out of the 
way, then there is the issue of which of IBM's many responses is the least 
harmful to the original project ... I usually take GO getin continue if it 
was JOBQ & Retry if it was interactive, but tell all my helpers to use Go 
because far too often they have done Retry on JOBQ after a bomb on say 
transaction 500 & it retries all 500 transactions resulting in duplicate 
postings.

I imagine there must be some AS/400 network & other places with examples of 
source code to reduce this kind of hassle for the end users.  For the 
interactive user, before trying to update some file that MIGHT have some 
conflict, I would want code from RPG program to detect whether or not that 
record is locked, then send user a better phrased message, including what 
response code to take after resolving the case of who is blocking access.

For the batch job in this scenario, what I ideally would like to end up with 
is.

1. Program wants to update a file with a record key, but in addition to the 
error message being IIML01 file it should include ITEM MASTER FILE which is 
how the users identify it, and it should also say which item # it is trying 
to update, in addition to wherever the system gets the record # which 
apparently is the only way to communicate through WRKOBJLCK.

2. Perhaps the program that wants to update a locked record calls a secondary 
program to handle the conflict, let's call it MSGAIDIIM if the conflict is 
with the IIM file, sending parameters identifying what item is to be updated 
& for what purpose.  The users might like to know the purpose so they can 
decide whether to bypass this update or if it is essential to get it done.

3. MSGAIDIIM program would
   a. unless this is the first iteration, check whether the end users had 
sent a certain GO message meaning, return control to the calling program with 
instructions to bypass that update.
   b. check whether access to that item is still blocked & if not, return 
control to the calling program with instructions to retry the update
   c. use some API (is there one?) to identify from WRKOBJLCK or some other 
LCK command precisely what program job etc. is blocking access to the desired 
item # record.
   d. only in the first iteration, create the message place that will be 
checked in later iterations to see if the end users sent the GO AROUND 
instruction.
   e. craft & send a message saying in effect that
the calling program / user / job needs to update item master item # for what 
purpose but
the blocking program / user / job needs to get out of the way first
can some users please arrange this, or 
do whatever 
to send the GO AROUND instruction
how
   this message should be sent to the user who owns the batch job that got 
stuck
   this message should be sent to the user of the job doing the blocking
   this message should be sent to certain computer department personnel or 
the work stations in the computer department
   the logic needs to avoid duplicates, like if the same user is on more than 
one list
   if we write such software then user or work station no longer exists on 
our system, the program needs to send a message to QSECOFR or QSYSMSG what 
message was tried to be sent by MSGAIDIIM to user or work station that no 
longer exists so that current computer department personnel know to make 
adjustments

   f if no response to first message there needs to be some loop & wait some 
reasonable time period - perhaps loop on "a & b" more rapidly than on sending 
message again, especially if the job doing the blocking changes.

Again, my need here is not urgent.
It is more that I am brainstorming some ideas.

MacWheel99@aol.com (Alister Wm Macintyre) (Al Mac)


+---
| 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
+---