Re: Counting users

Hi Everybody,

I ran into some difficulty responding to posts on the list. Everything was
coming back as Undeliverable. I finally resubscribed and everything seems
to be working again. I don't know how many times I keyed this explanation
of what I ended up doing, but here goes one more time.



 For dumb terminals, I think I'm going to look at port#/user/address and
 divide by 2.  So, if a user signs on twice to the same port, I'll count it
 once. But if there's three on the same port, I'll count it twice. This
 gives every user two sessions. (If they need more than that they can go to
 PC's and get as many as they want without a penalty).  And if a user signs
 on to 7 terminals on the same port, I'll only count it as 4 signons. So
 he'll get 3 freebies. But if he really needs dual sessions to be
 productive, as soon as they sign on to another session on one of those,
 the count will go up.


From: Carsten..
<<I noticed that you mentioned the list jobs API earlier in this thread -
for
<<this particular purpose the QEZLSGNU (List signed-on users) operational
<<assist API might be a better (and possibly faster) choice.

You are probably right, but I need more information for other requirements
to counting (group profile name and accounting code). Plus, I just used the
list jobs API because I had other programs that used it and I could clone
the code to get the thing up and running real fast. But it turns out the
speed on this is really impressive.  When I first started putting this
program together I was really concerned about speed. I thought I might need
to run the count program as a server program that was always running and
always kept a running count. Then I would only need to run the list jobs
API and actually start counting only when the number of signed on users
started approaching the max limit. So if the max limit was 50, the first 50
signons would not have to incur hardly any overhead associated with this
task.

But I thought I would just pull a preleminary program together to see if I
could get the necessary information to work and worry about the server side
code later. Much to my surprisse, the program runs in under 2 seconds!
Thats doing the following: Create a user space, create a user index, list
all active jobs into it, for each job that does not have a blank device
name run the retrieve device description API, for device types *DSP - add
the entry to the user index. When done adding all, retrieve the entries
from the user index and count em. All this in under 2 seconds per signon! I
was amazed actually. I am only getting 3 pieces of information from the
list so it runs extremely fast.

Again, many thanks to all who participated in this discussion. I was really
concerned about implementing a partial solution (not counting precisely,
which after all was the original task) but since this seems to be the
closest solution that anybody can come up with, I can live with it.


Ron Hawkins





Ron Hawkins

+---
| 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
+---