Concatenate 300 datasets as input to LMMLIST

coxdavi · New User Joined: 12 Mar 2009 Posts: 30 Location: usa

I have put together a REXX exec that uses library services to get a list of member information using the LMMLIST command with SAVE option so that the result is into a group dataset ( userid.groupid.MEMBERS ). This was done for one input PDS. I would like to get something together where I can take in up to 300 input datasets and run each one through the LM commands, putting the output into the same output dataset. So, for each dataset input, I would want the output information MOD'ed onto or added onto the end as each one is processed.

Another bit of info here is that the differentiating factor in each of the 300 datasets would be the high level which is the userid.

Any ideas on how best to concatenate this long list as input would be appreciated.

Thanks,
Dave

Pedro · Posted: Tue Nov 16, 2010 10:46 am

coxdavi · New User Joined: 12 Mar 2009 Posts: 30 Location: usa

Actually, I am looking to not have to run a REXX job 300 times. I am really looking for something where someone has already done the inside loop of LMINIT, LMOPEN, LMMLIST, LMFREE and LMCOPY where the LMMLIST is going to create a new list which would become the input to the LMCOPY command to copy DATA1 from the LMMLIST command to a new DATA2 dataid used as output in the LMCOPY command. The outer loop logic would have to somehow bring in each dataset which would become input to the LMMLIST command.

enrico-sorichetti · Posted: Tue Nov 16, 2010 10:35 pm

well good look

this is a help forum not a do it for me...
did You care to search the forums, I just checked and there are lots of code snippets

search for LMMLIST and start clicking on the results

coxdavi · New User Joined: 12 Mar 2009 Posts: 30 Location: usa

I don't need good luck. I need to find out how to handle 300 datasets which would be input to a LMMLIST and LMCOPY set of instructions. The fact is that I just found that if you specify ENQ(MOD) on the output dataid, you will not get a dataid defined for the output of the LMCOPY. Has to be ENQ(SHR). Back to the drawing board for how to store and mod onto the output dataset within LM. How do you think I found that out ? By looking around on the IBM LM pages, so please, if you can't offer suggestion that is constructive, please do not respond. On the other hand, if someone has constructive suggestion, please feel free. You might help more than just me.

Thanks.

daveporcelan · Posted: Wed Nov 17, 2010 1:00 am

Coxdavi,

Here is some constructive advise
1) Don't mess with enrico
2) Don't tell someone with over 6000 posts not to respond. They have earned the right to say anything they want. He has helped many, many people.
3) Enrico did give you three pieces on constructive advice
A) do not wait for someone to give you code
B) search the forum
C) search for LMMLIST
4) You already have the logic worked out in you original post. Continue on with that.
5) Don't mess with Enrico

enrico-sorichetti · Posted: Wed Nov 17, 2010 1:19 am

for today anybody can mess with me

just for today because I am in a very good mood, tomorrow I do not know

really the requirement is not so clear

I would split the process into 2 parts ...
1) get the dataset list and store it into a stem
2) process the stem

the advantage is that the dataset list can be provided in this way
1) by reading a unpatterned dataset list
2) building a dataset list with a pattern using the LMD**** stuff
or build the same using the IGGCSI interface

at this point once the dataset list is available

all depends on what process has to be done on each member
and I am desolated to say that it has not been explained too well

anyway
for the output I would not bother to use the LM*** services
if it is a PS dataset just ALLOCATE and write record by record using EXECIO

the only thing that is not clear is the output record content

once that the content is clear also the suggestions might be more useful

enrico-sorichetti · Posted: Wed Nov 17, 2010 1:37 am

for example getting a dataset list is not that difficult

enrico-sorichetti · Posted: Wed Nov 17, 2010 1:55 am

I suggested to search for LMMLIST to test the TS good will

I was sure that there was a working example... I WROTE IT ... right here
ibmmainframes.com/viewtopic.php?t=25947&highlight=

enrico-sorichetti · Posted: Wed Nov 17, 2010 2:16 am

and here another working sample up to the member list for a dataset list

coxdavi · New User Joined: 12 Mar 2009 Posts: 30 Location: usa

Thanks Enrico for the example. I already have the 300 datasets from LISTCAT and reduction manually. Not a big deal. The problem is introducing each dataset into the LM processing and being able to MOD onto the output. LMCOPY is out as the MOD doesn't work.

What is TS ?

dick scherrer · Posted: Wed Nov 17, 2010 2:28 am

Topic Starter

d

enrico-sorichetti · Posted: Wed Nov 17, 2010 2:32 am

as I said already use my snippet wher istead of using the LMD**** services to fill the DSLST. stem

is enough to use

coxdavi · New User Joined: 12 Mar 2009 Posts: 30 Location: usa

The output dataset should contain the list of members with stats for each dataset's members. I am trying to take in the first dataset input, get the list created, write it to the output, go back and get the next input dataset and repeat, only mod onto the output dataset. That was going to be the simple idea using the LMCOPY with LMINIT of ENQ(MOD) on the output dataset.

Thanks Dick.

enrico-sorichetti · Posted: Wed Nov 17, 2010 3:24 am

LMCOPY does not write single lines, it copies the content...

as i already said a few times the best approach is to write each line using

after the LMMLIST stats(YES)
just use

coxdavi · New User Joined: 12 Mar 2009 Posts: 30 Location: usa

Enrico, Thanks for your help today. Yes, I know that LMCOPY writes the whole shebang.

I was hoping to add on to the first set of members with the next and so on. Thanks again.

Dave

daveporcelan · Posted: Wed Nov 17, 2010 6:49 pm

coxdavi,

Have you noticed that Enrico has provided code on several occasions?

Have you also noticed, you have yet to show anything?

Very important piece of advice:

1) Show the code you are running, and how you invoke it (batch jcl for example).

2) Show the output it is currently producing (partial output is fine)

3) Show how you want the output to look.

This would make things easier for everyone.

If you would have done this from the get go, you would be done by now.

coxdavi · New User Joined: 12 Mar 2009 Posts: 30 Location: usa

I am done for now. We were able to get the MOD operation completed. This task will go on the shelf until I get some time to get back to it.

dick scherrer · Posted: Wed Nov 17, 2010 10:03 pm

Someone will be here when you get back to it. . .

d

coxdavi · New User Joined: 12 Mar 2009 Posts: 30 Location: usa

Thanks Dick. Much appreciated.