Re: Generic RPG File Handler for File I/O using OA

[Frank Kolmann <Frank.Kolmann@xxxxxxxxx>]

Hi Nate

I dont know Mahammad goals.
My goal was to learn UDDS coding on S38. I did actually use UDDS once or 
twice in production.
Plus I had a nice utility to view/update any file.

My later goals were to learn how to use IBM APIs, I did, but I never 
used in production.
Later I wanted to learn C, which I did, after learning C what I really 
gained was RPGILE was mostly a C based model.
After learning C many things became clearer, procedures , prototyping , 
pointers, modules and so on.

Lastly I wanted to  transition my utility to the latest version of RPG.  
SQL made the utility redundant, but it is a way to learn.
I failed because of the way NULL indicators are implemented, but learnt 
more  of Unix or is it C I/O functions to read SysI database.
Unix I/O is orders of magnitude more complex than RPG, not sure the 
complexity is worth the results.

Here I must mention my mentor, Scott Klement, without whose work I would 
never have progressed.
The S38 UDDS work was all mine own effort.

Finally my opinion on a Generic File Handler.  Yes I believe it is 
possible, but do you really want to get involved in the lower level 
detail of database access?
You end up redoing and taking over all the work the compilers do for 
you, and it is not really possible to do as well as the system 
developers have already done.
fwiw

Cheers
Frank



On 26/06/2022 9:31 am, Nathan Andelin wrote:
> Mohammad,
>
> I don't really understand your goal, nor the problem you may have with the
> traditional RPG operation codes for DB I/O. But you could review the C
> procedures, which are documented in the IBM i Knowledge Center, which are
> exported from service program QC2IO. You could do a search on function
> _Ropen, for example.
>
> The C routines are functionally comparable to the traditional RPG operation
> codes, except Library and File names are passed as parameters, so they can
> be resolved at runtime, as opposed to the RPG method, which requires
> Library and File names to be hard coded and resolved at compile time.
>
> Given the C routines, you could write utility procedures that resolve
> Library and File at runtime. For example, I wrote a generic procedure that
> checks for the existence of any record in any file, given a File name and
> Key value.
>
> The C routines perform a little faster than the RPG RLA alternatives.