System iNetwork

Don't want to ruin the lovefest, but I am not seeing core problems with the i5 being addressed.

On the hardware and pricing front, we don't know if customers will be allowed to buy market priced p5 hardware and then run $250 per user i5/OS on that hardware.

Object and field names are still limited to 10 characters in i5/OS.

ILE has not been improved since the 1990s and could be much improved. ILE programs should support reflection. With reflection the programmer does not need to supply a separate header file when compiling source code. .NET and Java don't need header files. RPG does. That is because ILE does not store procedure definitions and other metadata within the executable object.

i5/OS still has the 16 byte pointers it had back in 1978! Increase the size to 64 bytes. If the 5 byte segment, 3 byte offset structure of an i5/OS pointer had been increased, would the teraspace memory model have been necessary? Teraspace is great because it enables ILE programs to break the 16meg memory barrier. But they also complicate things because of their incompatibility with good things that SLS memory can do.

-Steve

Steve,

I think you're right, however the real issue is that IBM sucks at building GUI-based development tools. Sure SEU and PDM were state-of-the-art, in 1981 and 1988 respectively, but in 2007 most other platforms have generational GUI developer tools that are de facto standards (e.g., MS Visual Studio). Those companies throw good money at those tools and keep updating them. IBM goes nuts on things for 2 years or so and then walks away. And when user requests are made, they use the "poor me" line.

Sure other features like 64-byte pointers would be great. The discussion in the 1970's about pointers went "Pointers are never big enough, let's make them bigger than we'll ever need." Same mentality that went into date fields in the pre-Y2K era. But things changed, time continued and you really do need to update things.

Bottom line today, in my view the System i5 is to the System p, what the System/36 was to the AS/400. It is sad, disappointing, frustrating, and the opposite should be true; but, apparently it is true.

Everyone is entitled to their own opinion, but I don't think many programmers in i5 customer or ISV shops really care much about the size of pointers. . . . And if you just want to sell i5/OS on p5 hardware, you have effectively killed the i5 revenue stream and turned it into an HP3000 unless business partners actually started selling it on a zero margin platform . . . not likely. 515s are hard enough to make any profit on.

I agree that the 10 char field names change is long overdue. And I disagree that WDSc sucks. It isn't perfect, but then neither is Visual Studio (unless you are a Microsoft advocate). IBM's problem with funding better GUI tools for i5 is that they don't see a revenue stream for the tools from the platform because i5 essentially gives WDSc away with its tier based pricing (of course what the SWG guys don't get is that they get several million dollars in SWMA revenue each year), and their field sales guys don't see the revenue so they don't count it among the living. . . .

IBM tries to raise some revenue by moving features into the Advanced Edition, and then it gets hammered for charging for it. SWG ain't building tools for free. . . . O' the joys of yet another debate.... :o)

Wait a minute. Maybe I'm misunderstanding or getting hung up on semantics, but the basic software tier does not pay one cent toward WDSC. It is a purchased product paid for as part of the WDS (Websphere Development Studio) licensed product. Now that product (WDS) comes in a tiered license--depending on the tier, it can be relatively expensive. Perhaps that's what Doug means by "...essentially gives WDSC away with its tier based pricing...." But I wouldn't say that's essentially giving it away. He even notes that IBM receives "several million dollars in SWMA revenue each year."

And the whole thing about IBM getting hammered for charging for features by moving them into the Advanced Edition is ludicrous. IBM people are the ones who talked me into learning how to use CODE and CODE Designer. Then it was IBM's decision to designate that toolset as 'deprecated'. We're a very small shop--paying for the Advanced Edition would cost us over $20K to get the replacement for CODE Designer. IBM deserved that hammering. They've since back-peddled saying the components were only included in the Advanced Edition as a 'Technology Preview'. (I'll stick to calling it back-peddling because it doesn't fit with the comments I read that were made by George Farr at the System i developer conference (sometime early last spring?)) Then they really haven't addressed the other components that weren't moved to the Advanced Edition or the breaking out of EGL as an entirely separate product (so much for my investment of time on that, too). And yes, I'm aware of the 'free entitlements' that were available until July 31. However, the maintenance on those is only good for one year. I've tried in vain to find out what it'll cost to renew when the year is up.

I would have to agree that most i5 programmers don't care much about pointer size. I can do pretty much everything I need to with the current pointers.

Regarding the size of object names, I can give files and fields much larger names using SQL. IBM has done far trying to hide the fact that they don't want to do enhancements to DDS and the 'native' interface to the database--the focus is going to go heavily into the SQL interface. I'll have to allow that RPG doesn't natively support the longer field names and can't use the long available ALIAS names like COBOL has--but the fields in the database nor used in SQL need not be limited to 10 characters.

I would also have to agree that WDSC is far from being slapped with the label that it 'sucks.' I use it daily and while I can see that it needs improvement, it gets the job done far more efficiently than I could in SEU and PDM. I still believe I'm more efficient at building enterprise applications in the i5 environment than about anywhere else. Not to mention the reliability, stability, and security of the platform.

Some people need to get a grip. The issue for most developers has nothing to do with header records or pointer sizes. These are of minor concern compared to the overall direction IBM is taking with the RPG development environment. We need more emphasis on modernization to be able to write browser / client server GUI applications within the iSeries environment without having to resort to writing in JAVA, javascript understading CSS etc or CGI or migrating to ASNA and .Net. There appears to be no clear direction. Most iSeries shops have small development teams and tight deadlines and dont have time to try to determine the best methodology going forward when the experts have no clear ideas themselves.

Bob,

Thanks for the refreshing comments from a well known iseries 'anchorman' (Bob Cozzi). Your courage to criticize and speak out and bring in the open the weaknesses of this system (which to the general public is no secret at all), certainly gives a boost to your credibility.

At least the survival of the iseries may benifit better from those kind of introspective remarks instead of Ibm's endless we-are-the-best advertizing parades while everybody in this space runs in a covered-up fashion to the exits like Microsoft, SAP and Oracle.