Re: Dynamic Variable Names in RPGLE

On Tue, 2005-12-20 at 13:57 -0700, Mohammad Tanveer wrote:
> Any idea if we can use dynamic variable names in RPGLE?  
> 
> Something like this
> 
>  fldvalue = %value(&var1);
> 
> Once we chain to a file or create cursor program knows what fields are
> available.  Is there any API or Userspace stuff to get the value of the
> fields using a dynamic variable name?
> 
> 
> Thanks
> 
> 
> 


No.

However, you can roll your own to an extent.  In the case of a record
buffer, such as results from a chain or read, you can find out the
column name, type, offset and length at run time via DSPFFD to an
outfile, the list fields API, the SQL catalog, and others.

Once you know the column from positions and length for a particular
record, you could do

fldvalue = %subst(input_buffer: from_pos: length);

You will probably need some special handling for field types other than
alphanumeric.



Years ago, we had need to allow users to define their own validation
rules for any of about one hundred different fields contained on a set
of five screens.  Users needed to do simple validation against lists,
and also needed the capability to do rudimentary if-then-else logic, as
well as to compare one field to another field.  The rules change
frequently, and special cases are routinely added.

Initially, we though about generating a piece of code and compiling it
each time they changed the file maintenance, but we were concerned about
how cpu intensive that would be.  

Instead, we loaded the fields at run time into an array.  Once there, it
was easy to write generic code to compare (for example) array(15) to
array(10) or validate array(27) against a list.  

The only hard-coded portion of the program was that to move the fields
from the record into the array.  The rules were stored in a file, and
the 'engine' worked from the rules.


As usual, it depends on what problem you are trying to solve as to the
best way to go about it.  


Regards,
Rich