It's a DATA SET...or is it?

Nic Clouston · Posted: Thu Sep 15, 2016 8:06 pm

Everyone, please get your terminology right: they are not files but data sets

Girishb2,
Please use the code tags when presenting your data. Also, the records in your output data sets do not match your input sample data. How do you expect people to understand how you get from input to output if you do not show consistent samples?

RahulG31 · Active User Joined: 20 Dec 2014 Posts: 446 Location: USA

Nic,

I've been hearing this from a long time (the 'file' vs 'data set' thing). Probably, you would want to read the first line on this link:

https://www.ibm.com/support/knowledgecenter/zosbasics/com.ibm.zos.zconcepts/zconc_datasetintro.htm

It's not just anyone, it's IBM itself:

Akatsukami · Posted: Thu Sep 15, 2016 9:57 pm

Remember, though, that z/OS data sets are structured, with the labels and VTOC entries containing information that allow a program to use it without knowing that structure. To the left, files on toy computers (and even under zLinux) are unstructured streams of bytes to which the program (and programmer) must pay careful attention to interpret properly.

Arun Raj · Posted: Thu Sep 15, 2016 10:11 pm

Thanks Akatsukami. Still I would tend to agree to Rahul's point if we should really just ban the word "files" when we are talking about a simple processing logic. I just did a quick search on the word "file" and noticed thousands of posts from Frank, Kolusu, Nic Clouston and many other experts here where they all seem to have referred z/OS data sets as "files". So were those statements technically incorrect? I think they were n't. Sorry for deviating from the original topic here.

Click to Read : Allocating Files - Enterprise COBOL for z/OS 

Akatsukami · Posted: Thu Sep 15, 2016 10:24 pm

We all make mistakes. Even I have done so

Bill Woodger · Posted: Thu Sep 15, 2016 11:02 pm

Arun, the link you provided refers running Enterprise COBOL on USS using HFS.

The F in HFS is for "files", so obviously you'd expect "files" to be the correct terminology.

The F in zFS, on the other hand, is... oh rats. There goes that rock-solid argument.

I don't really mind using files/data sets interchangeably (I've just migrated to data sets as two separate words, so can finally admit to being so cavalier). However, my initial thoughts on seeing the word "file" are that perhaps the user of the word is not aware of the important distinction between "files" and "datasets", which is the "internal organisation (if any)".

Just using the correct word/phrase does not mean that the relevant differences (where multiple architectures are involved) are actually known, so a rote "file is not a data set" can mean nothing.

RahulG31 · Active User Joined: 20 Dec 2014 Posts: 446 Location: USA

Can somebody move the last couple of replies to a New topic 'data set vs files' ?

'data set' is an IBM preferred name for a kind of file which has a record organization. It's Not true that data set is Not a file and calling them 2 different things is Not correct (as this has been promoted for a long time on this forum).

@Bill: If you don't mind calling 'it' a data set or a file And most of the users don't mind it And IBM itself seems Not to mind, then I am against hearing that crap (it's not a file, it's a data set) all the time.

The link that I have posted clearly tells that data set refers to a file. That's it.

Most (Almost all) of the problems that users post, don't get solved by calling 'it' a data set instead of a file.

.

Bill Woodger · Posted: Thu Sep 15, 2016 11:36 pm

Don't read too much into one particular word of one particular article that happens to appear on an IBM website. There's stuff in the COBOL manuals which is plain wrong, for instance, is it just "right" because it is from "IBM"?

The context of that one word is interesting:

"The term data set refers to a file that contains one or more records."

OK, so a "something" which contains no records is... what? A file? If it is a file, then file and data set cannot be interchangeable. The sentence is meaningless, even though from IBM.

mistah kurtz · Posted: Thu Sep 15, 2016 11:54 pm

I was just wondering if its okay to be politically correct all the time?
Does it serve any real purpose?
Do we fail to understand a post(as an expert) if it refers to the term "file" and not "data set" if one does not mention RECFM and LRECL (which is required irrespective of the terminology)?
IMHO, bantering someone just because of this is I think quite discouraging!

RahulG31 · Active User Joined: 20 Dec 2014 Posts: 446 Location: USA

A 'data set' is a set of data. So, if there is No record, there is No data and there is No 'data set'. It is just an empty 'file'.

If you read Wikipedia, you'd find:

Robert Sample · Posted: Fri Sep 16, 2016 12:01 am

RahulG31, from the IBM Glossary of z/OS terms and abbreviations at www.ibm.com/support/knowledgecenter/zosbasics/com.ibm.zglossary.doc/zglossary.html :

Rohit Umarjikar · Posted: Fri Sep 16, 2016 12:06 am

I would take this way , Files are properly called data sets in MVS, Data set names (DSNs, mainframe term for filenames) or a data set (IBM preferred) or dataset is a computer file having a record organization. In all over DFSORT manuals it is termed as files too ,so is it then wrongly termed by the Great DFSORT or ICETOOL creator Frank?
So if someone asks which terminology is correct from MVS/IBM perspective then I would answer Data set. However, per my experience I hardly or in fact all the developers (20+ years of experience ) still say files all the times.
So as RahulG31 has pointed out rightly that stretching and insisting on naming Data sets instead of files to everyone isn't solve the problem in hand. So conclusion for me is 'Data Set' is much more accurate and prefered term to use but saying 'file' is not a crime.

Similarly, many of say's records in DB2 which is wrong and it should be called as Rows but very few of them follows the right terminology. So sometimes (not all the times) , it is a right approach to enforce the terminology constraints.

Arun Raj · Posted: Fri Sep 16, 2016 12:09 am

IMHO it is no serious crime like a generation vs version situation. Infact almost everyone of us have been using the words - file and data set interchangeably for quite a long time. So my question is if we ourselves are "not following terminology" and a bunch of manuals use the term 'files', then how can we ask others to stop using it?

Nic Clouston · Posted: Fri Sep 16, 2016 2:21 am

I admit tp having gone back and edited a post of mine that I made a few days ago and changed File to Data Set. I realise we all know what is meant when one writes 'file' but we are supposed to be professionals and our jobs are to get things right so we should strive to get our terminology right.

Recently I was reviewing my PL/1 tutorial from IBM. Here are a couple of extracts from the introduction to RECORD INPUT/OUTPUT...

o Data held externally to a program is held in data sets.
o External data is represented in a PL/1 programby a FILE. The actual data set to be processed is defined by JCL thus a program can be used to process different data sets without modification, so long as etc etc

What this means is that, for PL/1 anyway, a file is the DDName used and Data set is the actual data allocated to that DDname.

The manual did not go into whether it is Pea Ell One or Pea Ell Eye but within IBM I was told it stood for Programming Language One.

Akatsukami · Posted: Fri Sep 16, 2016 3:00 am

Willy Jensen · Posted: Fri Sep 16, 2016 1:56 pm

Dataset or file, shouldn't the context in which it is used make it obvious what type of operating system / file system / data structure we are talking about?
I must admit to be using 'file' more and more. And I really really don't care if someone uses 'file' instead of 'dataset' for a z/OS collection of data outside of a tape or a USS file system.
Just my 2 cents.