This thread discusses the Content article: IBM Reorganizes Itself to Reach Midmarket Customers

I'm sure I'll get a lot of disagreements, but here are my thoughts on this subject:

First, in response to Lee's comment "While I hate to say it, today's IBM may have lost the courage to listen to the mavericks within its ranks.": Considering the reorgs and "staffing actions" that have gone on in the past five years, I wonder how many mavericks are left at all.

Second, with respect to what should be done with the System i, I'm not sure what can be done. Today's situation in very different than when the AS/400 was developed. Compare the System i with the System p. The only real difference between the two is that one can run legacy RPG apps and is more expensive. In other words, pushing the System i involves selling RPG as a modern development platform. IBM basically has to find a way to make RPG more attractive than other alternatives which are either cheaper or provide significant productivity advantages or are simply more "cool". As someone who spent years working on RPG compilers at IBM, I'd like to think RPG has a good future. But we have to be realistic.

Cheers! Hans

Hans,

Realisticly, what features do you think would make RPG more modern, more productive, and cool? What would make it futuristic? _That_ would be cool! What's it lacking? In your wildest dreams, what would you like to see for RPG?

I'd like to think RPG has a good future too, and you're certainly in a position to have some interesting insights.


Hopefully this will spark a little life into this forum!


Tom D.

Tom Daly wrote:
Realisticly, what features do you think would make RPG more modern, more productive, and cool? What would make it futuristic? _That_ would be cool! What's it lacking? In your wildest dreams, what would you like to see for RPG? ...

Personally, I think we're way beyond that. I don't think putting half-dozen more developers (or more) on enhancing the language is going to make any difference in the long run. Besides, if you change the language too much at any one time, programmers complain about the training needed, or that they can't use the enhancements since they won't get to the current release for another 3 years, or the enhancements should have been done differently, or whatever.

Let me offer another angle on my analysis: The cheaper the System i gets in comparison to other systems, the more cost-effective it becomes to run interpreted OO languages on the system. That is, the more it becomes a commodity-priced solution, the less important RPG will be to the system. The less important RPG is, the less the need for the i5/OS and System i.

I know from direct personal experience how many companies are hiring System i programmers these days. In the GTA east area where I live, over the past 5 months there has been a total of one company hiring, and that one is in yet another declining business: cutting pre-recorded CD's and DVD's. I would be much more confident in the System i's future if there were more positions available.

The System i will survive for a while yet on the strength of its existing SMB applications. But over time, these will either drift over to commodity systems, or will get more competition from commodity-based solutions.

If you have a different take on the situation, let's hear it. I know there are others out there who see a brighter future.

Cheers! Hans

Setting aside these other issues and sticking just to the topic of the language, what would you like to see for RPG? IIRC OO is not on your hit parade. Namespaces? Aspects?


Tom D.

Tom Daly wrote:
Setting aside these other issues and sticking just to the topic of the language, what would you like to see for RPG? IIRC OO is not on your hit parade.

Off the top of my head, here are a few things:

First, complete free-form syntax. Come on guys, it's 2008! The fixed-form syntax alone is sufficient to rank RPG as a relic of the 1960's, and not a modern programming tool.

Next, I'd like to see a new implementation of character strings. The current CHAR VARYING is mired in performance problems. CHAR VARYING return values, although having a nice syntax, involve the copying of way too many bytes. Likewise, other character operations like catenation require too much movement of bytes to and from temps. RPG should have a new STRING type where a pointer addresses the string data in a string pool. Returning strings from procedures would involve returning just a pointer, not thousands of bytes. The STRING type should also handle all Unicode issues, which are also currently poorly addressed.

Third, pointers should be strongly typed. Many RPG programmers seem unwilling to use a more appropriate language like C for systems programming. In C, the strong pointer typing can help programmers find potential problems at the compile stage. Without strong typing, debugging pointer type mismatches at run-time can be very difficult and time-consuming.

Fourth, there should be some syntax for parameter polymorphism on procedure calls. That is, procedure call "foo(charvar)" would call one procedure and "foo(numvar)" would call another.

After that, there are lots of easy little things that could be done, such as %LOOKUP on array DS based on subfield, or SORT of array DS keyed on a subfield. Array element initialization should be easier.

One problem with this is that for some of these things, the old functionality still needs to be supported. This can be good or bad. For example, I know from experience that if you have the ability to bypass strong pointer type checking, you will. This results in a language that offers multiple different ways to achieve some coding problem depending on release. Other languages take a much different approach to R->R compatibility. Consider Perl 6 and Python 3: These languages are different than their predecessors. The owners of these languages have weighed the pros and cons, and made the decision to deprecate certain older language features. (I believe Perl 6 tries to recognize the older syntax, while the less drastic Python 3 offers a conversion tool.) Doing that for RPG would be very difficult since IBM's customers have come to expect strict R->R compatibility.

Cheers! Hans

Just to offer that RPG/ILE is the front-runner but aren't there also C and Cobol programmers out there on i5's, PL1 guys maybe?