Re: CL pgm and pointer variables

[Hans Boldt <boldt@xxxxxxxxxx>]

Steve Richter wrote:
> true. but that just begs the question of why a more generic MI CPY
> instruction could not be provided when the target is a data pointer. The
> important aspect of the data pointer is that it enables the procedure to do
> more work.
*If* the target were data pointer. But as far as I've seen, the
usefulness of data pointers in applications programming has yet to
be demonstrated.

> thanks.  I did not know of %decpos.  What about %DataType, esp since opdesc
> does not give data type as would be expected?
Yes, there are a lot of holes in current OPDESC support, and more
than 2 IBM labs are spinning their wheels over the issue.

But how would %DATATYPE solve that problem? When you pass a variable
by address, the only attribute of the variable passed that the
procedure really knows about is its address. You couldn't do
%DATATYPE on the parameter name and get the type of the variable
passed. We're not dealing with an OO language here, eh?

Could reference and const parameters be implemented by passing data
pointers instead? Well the first roadblock of course is that W-Code
just doesn't support data pointers, and I doubt we could convince
the back-end people to support them. But I must admit that the idea
is intriguing.

> have a good LOA,
I will! Well, my first priority will be to take care of household
matters while Sylvana takes care of the baby. But I do expect to
have plenty of time to work on my model train layout, as well as a
few other fun projects. I do not expect to be bored over the next 5
months!

Cheers!  Hans