RE: Calling QShell utilities directly - more thoughts

["Dennis Lovelady" <iseries@xxxxxxxxxxxx>]

> "If your utility program uses the ILE/C or ILE/C++ standard files for
> I/O, you can run your utility from either the qsh command line or the
> QCMD command line."

Yes, I have written many filters et al that can be (are) used from qsh.
Nothing new here.

> So I'm thinking that if IBM's utilities do the same, we can run them as
> I suggested.
>
> IBM Support did say that QShell relies heavily on environment variables
> and internal tables that are used for applying CCSIDs to the processing
> of commands. Still, the door is left open for direct calls, if those
> elements are not needed. Not sure how we would know that, except by
> experimentation at this time.

The problem with this is that "experimentation at this time" may not produce
equal results at a later time.  Scott essentially eluded to this fact.  We
have known (and it has been proven) time after time after time that if we
try to use some "feature" that we stumble upon, or some undocumented
approach seems to work, the results may change with the next PTF or the next
release or even the next attempt.

When I think of doing things like this, I have to ask myself: what are you
really saving, and what's the potential cost if it breaks down the road?

The first answer in this case is "almost nothing," and you have to answer
the second part for yourself.

Dennis

Dennis Lovelady
http://www.linkedin.com/in/dennislovelady
--
"It is only impossible until it is not."
        -- Captain Jean-Luc Picard