RE: RPG IV release levels and complexity

["Bob cozzi" <cozzi@xxxxxxxxx>]

Hans, 
The Python developer, since Python is an interpreted language, only has to
worry about their customer's upgrading to the new Python runtime. They don't
have to change operating system levels to use it. 
That is my point. The runtime should be available back to a reasonable
release level. I'm suggesting Version boundaries. If I have to buy or
subscribe and install a new version of the compiler with the new features,
I'm all for that. 
I see two directions here:
1) Let me pay you for the V5R3 compiler to install on my V5R1 machine. 
2) Let me pay you for the V5R3 compiler and require it to be used on the
V5R3 machine, but allow the runtime (compiled code) to work back to V5R1.

Option 1 is my preference, obviously because shops can take advantage of the
new features right away.
Option 2 is a compromise that I would agree with since it would allow
software to be written with the new features and yet still run on a few back
releases. 

As to the TGTRLS support, with the new technique you could just drop it. Why
worry about supporting/shipping TGTRLS V4R5 if you're not supporting it?
Should there ever be a move to 128-bit architecture and things like numeric
field lengths are different, then yes, an architecture-level "TGTRLS" type
thing would be viable, otherwise, the hell with it.

Bob Cozzi
Cozzi Consulting
www.rpgiv.com


-----Original Message-----
From: rpg400-l-bounces@xxxxxxxxxxxx [mailto:rpg400-l-bounces@xxxxxxxxxxxx]
On Behalf Of Hans Boldt
Sent: Tuesday, November 18, 2003 8:51 AM
To: rpg400-l@xxxxxxxxxxxx
Subject: Re: RPG IV release levels and complexity

Bob cozzi wrote:
> Is it me or is it just too difficult to track all the different subtle
> enhancements to the RPG IV compiler on every new release.
> 
> I?m consulting at a shop that is on V5R1 and another that is on V5R2.  Yet
> most people are still on V4Rx.
> 
> In reading articles and going to training I?m finding all of the
developers
> essentially abandoning RPG IV enhancements because they can never relay on
> their release being the one on which the feature is offered. 
> 
> Three contemporary examples (but things like this happen literally every
day
> as we move this shop to RPG IV):
> 
> Someone is all hot to use the new UPDATE %FIELDS() capability to replace
and
> EXCEPT opcode with Output specs? They spent ½ day trying to get that to
> work. Oops, sorry, that feature is in V5.2 not V5.1 :-(
> 
> Another one was trying to use qualified data structures. They?re at V5.1
so
> that?s great? but then when trying to take advantage of the Data
Structures
> as Arrays? the get compiler errors? Oops sorry, that?s a V5.2 feature, not
> V5.1.
> 
> Another one happened when on a V4R5 machine they tried to use qualified
data
> structures. Once again? Oops, that a V5.1 feature. 
> 
>  
> 
> There are but a very few of the daily occurrences going on in the shops
> where I?m consulting. 
> 
> Granted many people on this list don?t have these problems because they
try
> to stay in touch with the latest and greatest, but?
> 
>  
> 
> Is this a widespread issue or is it just me? I mean a few months ago I
> advocated that IBM either stop enhancing RPG IV on every release and only
do
> it once per Version. That way at least if you?re on V5, you have the
latest
> and greatest compiler features, or you can upgrade to it. As far back as
the
> year 2000 I met with the RPG compiler manager from IBM Toronto and
suggested
> the separate the compiler from the version and just ship Version X of RPG
IV
> which will run on OS/400 Version Y release Z and later.  For example, the
> ship version 2.0 of the RPG IV compiler and it runs on OS/400 Version 5.1
> and later. So it would work on 5.1, 5.2 and 5.3. This would give everyone
on
> V5.x (any release) the same set of features.  To me this is the only
> solution to this dilemma. 
> 
>>From where I see things, a major inhibitor to getting shops to move to RPG
> IV is when they set out a few recon programmers to find out if its
feasible,
> you want that frustration level as low as possible. The way the compiler
is
> today, that is just not possible.
> 
> And forget that argument about why it can?t be done, who cares why it
can?t
> be done. I want to know if it is I a problem for us developers or if this
is
> an isolated situation. 
> 

I really don't know why, but I feel compelled to put in my two cents 
worth in this thread.

First, this "problem" is something that afflicts programmers in pretty 
much all languages. For example, the Python programmer who uses version 
2.2 is out of luck if he wants to use the new CSV file classes, which 
were introduced in version 2.3. He too grumbles before he (and possibly 
all his clients) upgrades.

Second, the suggestion was made that enhancements should be PTF'ed to 
previous (and possibly unsupported) versions. Do you know what's 
involved in putting out a PTF? Say we have to PTF a problem back to 
V5R1. First, the V5R1 code needs to be changed and tested, and PTF's 
have to be built for V5R1, the TGTRLS(V5R1M0) compiler for V5R2, and the 
TGTRLS(V5R1M0) compiler for V5R3. Then, the code change has to be 
propogated to the V5R2 compiler and tested, and PTF's built for that and 
the TGTRLS(V5R2M0) compiler for V5R3. And then, the change is propogated 
to V5R3. In most cases, the same change is made to each version, but 
sometimes the problem needs to be solved differently in each version.

My point here is that if we PTF'ed enhancements, we'd be spending all 
our time in building PTF's with no time left over for developing the 
enhancements. :-/

Third, since the RPG language is very much tied to the operating system, 
I doubt that we could ever fully divorce the compiler from the O/S 
release schedule. In V5R3, at least one enhancement definitely depends 
on system support.

Fourth, enhancements only on version boundaries? I don't see how that 
would fly. Even on the current release schedules, changes don't happen 
fast enough for many programmers. And as far as I can tell, there really 
isn't any system to the planning for new versions.

Finally, in my opinion, the pace of enhancements in the past few 
releases has indeed been too aggressive. Not only are some (many?) RPG 
programmers having trouble keeping up, but we in compiler development 
have been bitten off more than we can chew in several cases. That, and 
the fact that the ILE RPG compiler code is getting rather "mature", 
means that I believe the pace of improvement to the language will slow 
down in future releases, perhaps allowing programmers the chance to 
catch up somewhat.

Cheers! Hans

(Sigh - I need another LOA! ;-)


_______________________________________________
This is the RPG programming on the AS400 / iSeries (RPG400-L) mailing list
To post a message email: RPG400-L@xxxxxxxxxxxx
To subscribe, unsubscribe, or change list options,
visit: http://lists.midrange.com/mailman/listinfo/rpg400-l
or email: RPG400-L-request@xxxxxxxxxxxx
Before posting, please take a moment to review the archives
at http://archive.midrange.com/rpg400-l.