File definition in RPG all free

While I have been working with the new RPG all free (released in November 2013) I have unable to find examples comparing the pre-change to the post-change RPG code. Therefore, I thought I would give some examples of how to define files here.

When I first heard that RPG had gone completely free I was concerned that we could now intermix everything. This scared me as I had vision of people intermixing file definitions with the equivalent of calculations, etc. Fortunately what I have found is while you can intermix the new definition statements, you cannot insert them into the equivalent of the "calculation specifications".

In the move to RPG all free the following specifications have been replaced:

• Control (header) specifications with ctl-opt
• File specifications with dcl-f
• Definition specifications with dcl-s and dcl-ds
• Procedure specifications with dcl-pr, dcl-pi, dcl-proc, end-proc
• Input and Output specifications remain in fixed format, I think this is a big hint from IBM that we should no longer be using these specifications.

At the time I am writing this post to be able to use the RPG all free your IBM i server needs to be running 7.1 and have PTF SI51094, or a later superseding PTF, loaded. If you do not know if you have this PTF loaded read How to tell if you have the PTF for RPG all free.

What I am going to do is give an example of a File specification and then its equivalent in free. I am not going to cover every possibility of how to define a file, these are the most common ways I define files.

All file definitions start the declare file word, dcl-f, and end with a semicolon, ;. I have found that it is not truly free format. There are certain keywords that must occur before other. If you get a compile or syntax errors try rearranging the keywords.

The first example is very simple, a data file is being defined as input and without a key. As this as this is a data file I would use the disk keyword. There is now a usage keyword that defines how the file is used, which in this example is just for input.

FFilename++IPEASF.....L.....A.Device+.Keywords FTESTPF IF E DISK dcl-f TESTPF disk usage(*input) ;

Both the disk and usage(*input) are optional. I could code the file without them and they are assumed.

dcl-f TESTPF ;

Now let’s make this a bit more complicated. In this example I still only want the file for input, I want to access the file as keyed so I use the keyed keyword, and I am going to use the prefix keyword. Notice how I have also not used the disk keyword.

FFilename++IPEASF.....L.....A.Device+.Keywords FTESTPF IF E K DISK prefix(a_) dcl-f TESTPF usage(*input) keyed prefix(a_) ;

Now I want to update the file so the usage becomes *update, and to rename the record format I am using the rename keyword that was available pre-TR7. Notice how the dcl-f code now takes two lines.

FFilename++IPEASF.....L.....A.Device+.Keywords FTESTPF UF E K DISK rename(TESTPFR:NEWNAME) dcl-f TESTPF usage(*update) keyed rename(TESTPFR:NEWNAME) ;

Having used a file for input and update what about for just output, usage is *output.

Why do I do that? Even though the file itself is keyed no-one has ever given me a good reason to define a file in a RPG program as keyed if I am just writing to it.

FFilename++IPEASF.....L.....A.Device+.Keywords FTESTPF O E DISK dcl-f TESTPF usage(*output) ;

If I have a file that I want to read (input) and write to (output), without updates, I would code the usage for both *input and *output. Notice how they are separated with a colon, :.

FFilename++IPEASF.....L.....A.Device+.Keywords FTESTPF IF A E K DISK dcl-f TESTPF usage(*input:*output) keyed ;

In this example I want to be able to read (input), write to (output), without update. The usage would just be coded for *update and *output, it assumes that input is needed as part of the update.

FFilename++IPEASF.....L.....A.Device+.Keywords FTESTPF UF A E K DISK dcl-f TESTPF keyed usage(*update:*output) ;

There is one addition to the new, a new type of usage. If you want to delete records from a file its usage must include *delete. In this example I want to read (input) from the file and then delete records. I do not have to give *input as that is assumed with the *delete.

FFilename++IPEASF.....L.....A.Device+.Keywords FTESTPF UF E K DISK dcl-f TESTPF keyed usage(*delete) ;

And what about an update with write (output) and delete.

FFilename++IPEASF.....L.....A.Device+.Keywords FTESTPF UF A E K DISK dcl-f TESTPF keyed usage(*update:*output:*delete) ;

After data files, physical files, logical files, SQL tables and SQL indexes, the next most common type of file I use is a printer file. As this is not a disk file I do have to give the file type, printer.

FFilename++IPEASF.....L.....A.Device+.Keywords FTESTPRTF O E PRINTER dcl-f TESTPRTF printer ;

As with printer files I have to give the file type for a display file. As I am using the display file for both input and output I do not have to give the usage as input and output is assumed. I have also used the indicator data structure keyword, indds, and as the display contains a subfile the sfile keyword.

FFilename++IPEASF.....L.....A.Device+.Keywords FTESTDSPF CF E WORKSTN indds(IndDs) F sfile(SFL01:Z1RRN) dcl-f TESTDPSF workstn indds(IndDs) sfile(SFL01:Z1RRN) ;

I know that the above examples do not cover every type of file and scenario. I hope that this gives you a good basis of how to code the dcl-f in your own work.

You can use PDM for coding RPG all free, I did for this post to show that it was possible. Before you do you need to turn the syntax checking off. I explained how to do so in the post TR7 available today.

 

You should read my other posts introducing all free RPG:

• Defining variables in RPG all free
• Defining Procedures in RPG all free
• Mixing it up with RPG all free