How to dynamically decide which one to terse and not?

javen777 · New User Joined: 06 Mar 2015 Posts: 31 Location: china

Dear friends,

Currently i am going to implement a new function in system.
Which is to receive upstream data files and process them necessarily.

New Function Process Flow:
1.upstream files are generated in our MVS system.
2. Once each file is generated, it will trigger 1 OPC automatically to :
1) unterse the file if it was tersed, if not tersed then skip the unterse step.
2) After 1) , rename the file itself to our naming standard.

upstream file names are like:
FDCAP.WE.RECEIVED.W401F01.TERSED.FTP (terse file has TERSED in qualifiers)
FDCAP.PROD.D0000000.EXTFEED.G101F05.FTP
FDCAP.I.P1.FMSDC.G101.FTP
FDCAP.WS.W401F03.FEED

our naming standard:
TFMS.F.DD.USS.<SYSIDSRCID>

for example:
FDCAP.PROD.D0000000.EXTFEED.G101F05.FTP - >TFMS.F.DD.USS.G101F05
FDCAP.WS.W401F03.FEED ->TFMS.F.DD.USS.W401F03

Could you guys plz help on this?

Thank you very much

enrico-sorichetti · Posted: Wed Dec 19, 2018 9:12 am

amaterse was never supposed to be a general use use utility,
it is not even described in the utilities reference manual

it is a service aid used to exchange information with the IBM support organisation

wiser to investigate a different approach

javen777 · New User Joined: 06 Mar 2015 Posts: 31 Location: china

steve-myers · Posted: Wed Dec 19, 2018 10:54 am

About the only semi reliable way to deduce if a file represents a class of main frame data set is to look at its data length.

Multiple of -

1024 TERSE data set
4160 SYS1.DUMPxx data set

javen777 · New User Joined: 06 Mar 2015 Posts: 31 Location: china

steve-myers · Posted: Wed Dec 19, 2018 1:03 pm

I did not say the method is 100% reliable. I just said it's the only method I know of that can serve as a guide. Mainframe data sets containing tersed data are always RECFM=FB, LRECL=1024

prino · Posted: Wed Dec 19, 2018 1:24 pm

There is no, no, no, no (repeated ad infinitum) 100% reliable way other than to use AMATERSE to (try to) UNPACK the dataset. Just as I can create a dataset that seems to be ISPF PACKED by editing the first few bytes in HEX ON (and making it un-openable with the ISPF editor once it's saved), I can do the same with an FB/LRECL(1024) dataset, where the first 8(?) bytes of the first record contain information AMATERSE needs to properly UNPACK the dataset.

And if a Terse-Bomb can be made, UNPACKing an unknown file that looks like a TERSE'd one is not a good thing...

javen777 · New User Joined: 06 Mar 2015 Posts: 31 Location: china

prino · Posted: Wed Dec 19, 2018 1:42 pm

Are you stupid, or just pretending? What can't you understand about my previous reply that tells you that only, only, only (repeat ad infinitum) AMATERSE can decide that a file is validly tersed.

And what is "a REXX"? The same as "a COBOL", or "a PL/I"?

Marso · Posted: Thu Dec 20, 2018 9:22 pm

You can know if a data set is tersed only by looking at its name:

Willy Jensen · Posted: Fri Dec 21, 2018 4:26 am

No it wouldn't, the host name depends on the transmit command. I have uploaded AMATERSEd files to various zos dataset names.
What you can do is (as suggested above) is to
1. If recfm/lrecl if F(B)/1204 run a AMATERSE step to a dummy output dataset.
2. If RC=0 then run a AMATERSE step to a permanent output dataset.
As stated several times above there is no way you can determine if a dataset is AMATERSEd from just looking at the dataset specs and the data, period.

Willy Jensen · Posted: Fri Dec 21, 2018 4:48 am

Come to think about it, skip the test for recfm/lrecl, just run an AMATERSE step to a dummy output dataset and if RC=0 then run a AMATERSE step to a permanent output dataset.

javen777 · New User Joined: 06 Mar 2015 Posts: 31 Location: china

steve-myers · Posted: Fri Dec 21, 2018 9:45 am

Willy Jensen · Posted: Fri Dec 21, 2018 3:13 pm

I did a quick test and found that AMATERSE would very quickly determine if the dataset specs were wrong, or the start of the dataset was not right. In fact I would expect the AMATERSE check to be faster than starting a REXX program to check the dataset specs. Then again, I could be wrong. One advantage with a REXX check is that you could run the AMATERSE from the REXX, thus eliminating the 2nd step.

Willy Jensen · Posted: Fri Dec 21, 2018 3:39 pm

Sorry, I see what you mean now. Well, the 2nd step could be an IEBGENER from the temporary dataset, which would be much faster than AMATERSE. My suggestion was because it had earlier been mentioned that a FB/1024 dataset is not neccessarily an AMATERSEd datataset, so you would have to run AMATERSE to determine if it is.

enrico-sorichetti · Posted: Fri Dec 21, 2018 5:05 pm

the TS already clarified that ...