Prior release support (was: open source rpg compiler)

["Jon Paris" <Jon.Paris@xxxxxxxxxxxxxx>]

 >> It's not impossible by any means but it requires a significant change in
the mindset of existing iSeries software consumers.  Today, if I compile for
TGTRLS(V5R2), I only need to tell my customers that this software runs on
V5R2.

It also requires a significant change in IBM's thinking - and I'm not just
talking about the compiler team.

Assume for the moment that Hans, Barbara and the rest of the Toronto team
decide tomorrow that their next release will be backward compatible.  How
far back can it go?  Well - unless there is a significant change in IBM's
(not just Toronto's) thinking - the likelihood is that V5R1 would be the
absolute earliest release you could ever hope for.  Why?  because IBM does
not issue PTFs for out of service releases.  So one of two things would have
to happen - either they change that policy and start issuing PTFs for out of
service releases (an event that might possibly occur about two years after
hell freezes over) or alternatively the Labs start issuing fixes
independently of the PTF mechanism.  Being realistic, that won't happen
either - it just isn't practical.

Besides, the vast majority of people who have V4R5 or later are updating
their releases, it is the ones who never moved off V3R6/7 (or are still on
white boxes) that are the problem and there is no way the compilers could
ever target that far back.

As to open sourcing the compiler - I can't see it happening.  Even if
Toronto would give up the source to the compiler, it would require that the
details of w-code and/or MI prime be made available.  Since these are
proprietary and require the agreement of multiple groups within IBM ....
well let's just say I am not holding my breath.  The only practical option
would be to build a new compiler that generated (say) C code.

Frankly I think we'd all be better served by focusing on: 1) open source
projects for the development of a comprehensive library of procedures and
functions and 2) Instead of ranting about the compiler, focus our energies
on lobbying IBM to fix the bloody operational descriptors so that we can
write the kind of universal routines (like Joe's %Move) that we all dream of
building.

To paraphrase the comments that an IBM manager once wrote on my appraisal
"We all need to be a bit more selective in our choice of windmills to tilt
at".

Jon Paris
Partner400
www.Partner400.com