Re: CL parameters over 32 length

On 7/5/2018 11:25 AM, Jon Paris wrote:

Item 1 in your list won't really help Buck - it has been a long time since we were limited to 64K and I think a 16Mb temp would for sure cause issues!

I sort of wrote that tongue in cheek, but it sets the table for a better
idea.

Item 2 (full descriptor support) has long been requested but not implemented. In most cases you can at least now use the PCML support to embed the definitions in the program - and of course there's an API to retrieve those. Problem is that the PCML support is also incomplete.

I so, so, SO wish Rochester would do something about this. It's one of
the real gaps in API / tooling support that I really miss.

Item 3 is an interesting idea - but I don't see it helping much in practice. Problem as I see it is that it presumes that the person making the call knows exactly what the callee expects - and in most cases had the programmer realized that he wouldn't create the problem in the first place!

I myself mostly field questions about 'garbage in my parameter' during a
testing session, when a developer wants to run one program in isolation
from the rest of a big CL stream. For ad hoc requests like this, it
might be more palatable to have CALL xxx ARGDEF() than writing a *CMD.