System iNetwork

Chris,

Would one of the experts you have in mind be Greg Hintermeister from IBM? He has written extensively about his team's work on IBM's iSeries Navigator and Director product lines, and has just written a new article on Navigator for i5/OS which provides a new Web based interface for managing the System i, which will be released in V6R1.

I ask for the benefit of your readers who are more closely aligned with perspective #2. Although I'm personally more aligned with perspective #1, it appears to me that Greg Hintermeister would fit the description of a strong-willed champion within IBM who also presents a clear vision of IBM's GUI directions with respect to i5/OS systems management.

Those readers who want IBM to anoint one GUI interface for System i business applications, also want IBM to use the same interface for i5/OS systems management.

The dilemma perhaps is that it also seems likely that Greg Hintermeister would catch a lot of flack from those who align themselves with perspective #1 *[CM Note: See Nathan's follow-up post -- he meant #2 in this paragraph] because IBM has largely chosen to develop in C and Java, and use Windows thick client GUI interfaces and standards-based J2EE interfaces, while those who align themselves with perspective #1 appear to want a native (ILE based) interface.

Perhaps the best way to describe some of your readers is with an oxymoron such as recalcitrant-loyalist; people who are fiercely loyal to IBM and consistently refuse to accept solutions from any other source, but at the same time stubbornly refuse to adopt the development tools, languages, interfaces, and design patterns that IBM has adopted.

[Note From Chris: Great suggestion, Nathan! It's been a couple of years since I've talked to Greg, and your characterization of the problem speaks to some potential sticking points with IBM's goals with other solutions and how it relates to the System i . . . which I think only IBM will be able to articulate much further.]

I think you have to first define your terms. Is the GUI being discussed intended to a) augment, enhance and build on existing interactive 5250 apps? Or is it b), something used from the browser or client PC that integrates barely or not at all with green screen code?

When IBM and industry experts propose and advance GUIs for the System i, it is always a GUI in the "b" group. B group GUIs fail in the system i market because most of the programmers have green screen skills, maintain greeen screen apps, and are looking for 5250 GUI extensions, not GUI replacements.

-Steve

I think you should check with the third party software providers. Ask them about the architecture that they support with their products. I think the fact that some of the products use CGI, and the response that IBM received with their proposal to drop support for CGIDEV, should tell you that prospective #2 is technically viable.

The bottom line is the bottom line. I have built client/server, n-tier client/server, and CGI applications on the System i. The time required usually meets or exceeds the time required for the original product build.

This is acedemic for me. My company dropped the System i after decades of use, and my job is now supporting Lotus and JSP applications.

My first post in this blog had a couple mistakes in the fourth paragraph. I meant to say that Greg Hintermeister could catch flack from proponents of perspective #2 - the people who want IBM to anoint and stick with one native GUI for the platform (for the OS as well as business applications).

Products like Webfacing, Newlook, and PSC/400 build on DDS and the 5250 interface, but I think most experts would agree that type of interface is an interim solution, at best. Others would say they just add eye candy but no real value, and the overhead and limitations actually detract from 5250.

Users want a better user interface - not just an extension of DDS. They want to be able to easily swap between multiple concurrent running applications with a single mouse click, splitter bars that divide the screen vertically or horizontally, drag and drop capability, optional help text to appear while hovering over an input field, the currently selected record in the list to be highlighted, and for the highlight bar to move from row to row when the arrow keys are pressed, and the record panel to be updated when a new record is selected from the list, asynchronous chat capability, options to export the data on the screen to an Excel spreadsheet, tabs on the screen where the state of an active tab is preserved when navigating between tabs, drill down links that enable users to navigate along any point in the hierarchy with a single click, and so forth - much of which is beyond the capability of the 5250 paradigm.

I think IBM basically understands the feasibility of offering a native GUI extension to the 5250 interface, but also realizes that it wouldn't be much different than the interim solutions mentioned above.

I want to point back to an excellent summary by a contributor only known as "MC" in a previous "Maxed Out" blog.

In it he reminds us that both Microsoft and Oracle had worked out their primary IDE and GUI frameworks years ago. Their committed client/developer base waits for each release with great anticipation.

We have had none of this in the i-World and Chris and others are right that we have fragmentation in third-party solutions that my Employer will not buy. And in addition to a conservative developer base, we have also had employers who switched to Microsoft, for example, because the company placed its entire prestige behind their IDE and GUI solutions. So let's not completely blame programmers who would not adopt one of the many, many 3rd-party solutions.

There was a day when COMMON resolutions were excitedly read and the responses from IBM were awaited with great anticipation. That day is long gone and dropping that process was a strategic blunder by both IBM and COMMON. It is with great regret I write of this. It is a great loss to me personally and I think our community suffered when the forward momentum was lost.

Nathan is absolutely correct. We need a interface mechanism that includes the full capabilities that a PC can deliver including custom built widgets. But it needs to provide a transition mechanism for those that haven't jumped yet that is actually usable and plain HTML web pages are not even close. It has to allow an in house programmer to build a complete app like he/she use to. It has to allow the easy use and reuse of other programmers code like the way we can currently call another vendor's program from our program to display information. IBM needs to use it to operate the machine. It has to be a UI that everyone will WANT to adopt because it is fast and powerful.

IBM doesn't have to build, just commit to it so there is enough at stake that what ever it is will live for many years. Unlike about half of the GUI vendors in the i5 arena over the last decade.

When IBM, vendors and inhouse all use the same UI then we gain the ability to buy packaged software and connect it to our other packages and in house software.

Mind you this is all for in house apps. HTML web pages are fine for public access on company web sites. HTML is just not suitable for day to day internal business. The users want, and I want responsive, interactive, intuitive interfaces. A fully usable web 2.0 solution would fit this bill.

A solution that fills these requirements will return the economic efficiencies that made this box so popular.

IBM cannot say they are courting SMBs and in the same breath say things like build the program and then let the web designer make it pretty. Who? What web designer? There is just me, Fred, and Mary the data entry clerk here. I shouldn't need a web designer to build business apps.

I will let you in on a little phsycological secret to UI design. If the buttons and boxes have square corners it is thought of as old. Like the word legacy. No matter how pretty the color scheme. You can't sell old software.

If you haven't seen Microsoft's XAML for developing a user interface check it out. Very easy to specify the entire user interface in a few lines of code / per screen. XAML is part of WPF/E Microsofts attempt at web 2.0. The /E stands for everywhere as in non-windows platforms.

Adobe also has FLEX a layer on top of Flash that uses MXML to specify a screen and is also very functional. Again with just a few lines of code the entire screen can be defined. FLEX exists now and runs on Windows, Linux, Unix, OSX, etc. Flex also uses Eclipse as the editor.

DOJO has been out there a while but uses java script to define everything. The entire platform is java script. No comparison to XAML or MXML.

WEB 2.0 is coming and will make ubiquitous apps. Can the i5 play in that market? Will IBM make this powerful unstopable box become the Web 2.0 killer machine?

John mentions the need for a trademark GUI IDE from IBM...that fits for the System i. I agree that it's required, and I think I might have some insight into why it didn't happen...and a suggestion for making it happen now.

I recently met a guy who used to work on the RPG compiler in the Toronto Lab. He said that when IBM tried unsuccesfuly a couple decades ago to bring all the IBM developers onto Case tools, and other modern (back then) development environments, they failed. Nobody would bite.

He said the mainframe programmers would only use COBOL and the AIX programmers would only use C and the AS/400 programmers would only use RPG.

So IBM sucked in their collective gut and said "If they won't do what we want, then we'll build what they will use...and still make a pile of cash." Or something like that.

So perhaps, we are asking too much of IBM. They are only did what they thought we were telling them to do. And it seems that they (foolishly?) determined what we wanted by watching what we did (or didn't) do.

Maybe, if we all stopped writing green screen programs and started writing Web 2.0 apps on Windows servers, they would say "Hey-looks like they want WUI (Web 2.0 UI) - let's build that into the next ix/OS!"

I guess I'm coming to the conclusion that it's up to us - the developer community - to turn the tide at IBM. If we all pick the standard UI and stick with it, then they will make that the standard UI on the System ix and sell it to us.

Any suggestions on how we can do that?

Why not talk to Tim Massaro who is currently an advisory programmer on the Application Development Solutions Team at IBM, Rochester.

I just read an article he wrote outlining how to use Ruby and Ruby on Rails. He also mentions PHP, which I'm told can be used to walk on the moon, if required.

Maybe he can shed some light on how to get IBM to catch a fire on something that we can all use.

When the AS/400 was introduced in 1988, it took off like California wild fires fanned by Santa Ana winds, and I think most application developers of that era (including me) would agree that its success was largely due to the highly productive languages and tools for developing interactive applications, which were tightly integrated with the OS. So I understand the emotional attachment and business proposition for asking IBM to do the same with a GUI interface.

However it's totally off-base to claim or imply that Microsoft and Oracle followed that strategy. I've spent years developing under Microsoft and have quite a bit of experience with Oracle and their success is definitely not due to their unified development or runtime environments or even their GUI interfaces.

I acknowledge that Windows is based on a GUI interface. Indeed the highest priority task under Windows is the one that monitors and responds to keyboard and mouse movement, but that's also one of the biggest detriments to using Windows for multi-user, multi-tasking, server-based operations. It would definitely be bad if I5/OS were to evolve in that direction.

Microsoft's success in the server arena is largely based on leveraging their monopoly in the desktop arena, but Microsoft's history is the antithesis of unified-integrated development environments. My primary reason for switching from Windows to I5/OS development was because Microsoft tends to redefine and redesign their development paradigm every 3-5 years. Your applications tend to become obsolete after 3-5 years.

Microsoft's recent consolidation under .Net is quite comparable to IBM's recent consolidation under J2EE, both of which don't compare very well with the native interfaces or architecture of I5/OS. In fact, both .Net and J2EE are distributed architectures with tons of frameworks from 3rd party providers, which is completely contrary to the call from traditional I5/OS developers for IBM to anoint just one GUI interface for both OS and business applications and stick with it.

Certainly someone in IBM with a lot of insight is Robert Cancilla of IBM Rational Tools and i/z Strategy.

One of his quotes is, "As of March 1, 2007, all System i compilers, ADTS (SEU, PDM, SDA, RLU, DFU), WDSC (standard and advanced edition) and WDHT (HATS) were transferred from WebSphere to Rational".

He has a lot to say about EGL and a possible career path for a good number of us.



Thanks!

--John

Nathan - it seems like you've thought a lot about this - definitely much more than I have. But now I'm stuck. If I'm in front of an executive team trying to present the System i as a reliable machine that they can use for in-house development, am I supposed to position it as capable of playing with most GUI frameworks (even though it may need to be replaced in 5 years), or as a machine that comes with a "unified-integrated development environment" (that nobody likes to look at)?

And why can't we have both?