From Gen Server to Web Service in 30 minutes

August 29, 2007

Following on from my last post - it was refreshingly easy to take a Gen procedure step that I had written in 1999 (and never changed - but still working and using action blocks written by other people in 1997 and 199 8) and get a Java proxy built out of it and apply the Web Service Wizard and deploy (ok - in a “proof of concept” environment) a simple Web Service.

Additionally, aside from some configuration issues, the service was generated in total, in around 30 minutes!

“If it sounds too good to be true - then it probably is”… is what people normally say about these type of claims, but in this case, I have to support the fact that it IS good, and it IS true !

The Java Proxy just worked, out of the box (see last post!!!) and the Web Service Wizard gave me some WSDL that, again, simply worked with no adjustment. We formed a view that to publish an existing Gen procedure step would require a small amount of rework to adjust the views presented to the outside world, a bit of repackaging and hey-presto, there’s your web service done!

The benefit of all of this of course is that the business logic is contained in one model, and in our case, deployed on the same set of boxes. Taking the alternative approach of building stuff from scratch would have taken months. Surely this is the ultimate in “ROI” - returning the investment by reusing whole systems has gotta be the way to go……

add to del.icio.us Add to Blinkslist add to furlDigg it add to ma.gnolia Stumble It add to simpy seed the vine TailRank


Out of the box

August 26, 2007

I recently did some work with Gen in getting a web service generated with the Java Proxy, and then using the Web Services Wizard to take that code and generate a Web Service from it (i.e. give me the WSDL).

The process is good, and simple, but it appears that there is no command-line interface to the Wizard. To be effective, I believe that it should be part of the toolset and be selectable on the generation options window and (depending on deployment options) generate a Java Proxy or an EJB as the basis for the WSDL, and have those selections drivable from a command line for automated builds.

Hopefully in upcoming releases of the toolset this may be a possibility.

add to del.icio.us Add to Blinkslist add to furlDigg it add to ma.gnolia Stumble It add to simpy seed the vine TailRank


Development - top-down, inside-out, middle out……?

August 19, 2007

I recently came across this post, from the development log for Microsoft’s Visual Studio tools - I know it’s dated 2005, but some of the points are relevant. He believes that the approach that IEF took -which was a totally model-driven approach (he was on the design team in the late 1980’s/early 1990’s) could be successul, but now a “nimbler way forward” is better.

I believe that he says this because he is an advocate of Domain Specific Languages, and that for each specific role in a large enterprise, a specific DSL should be chosen, and that they be integrated by standards-based connectivity. Gen is, as he puts it, entirely self-contained. He advocates replacing the model with source code control.

I would put it a different way.

Source code control, to me is a backward step, and model-based development is a forward step ! I do agree with what he says, though on standards-based connectivity between disparate technologies. DSL’s have been around for years - haven’t they ? We used assembler for the low-level, fast, intricate stuff,with COBOL for the bulk data-processing elements of the system.

What was missing then was the standards based connectivity between parts of systems. Now we have that, all is well, since as long as Gen can adhere to the standards, then Gen can be part of the “new world”.

Gen (in my mind) could be considered a DSL, BEST SUITED for enterprise-scale, reliable data processing operations.

In addition, Bill thinks that a code and model approach is also required, presumably with a two-way interaction between the model and the code and the code and the model. I don’t believe this in a Gen environment, since Gen’s strength is NOT allowing the developers to see the code, and allowing them to violate model constraints - also, having direct access to the code assumes a specific target platform - how many of Microsoft’s tools can deploy the same piece of business logic on a mainframe, PC or Unix server ???.

Microsoft’s way forward with these tools assumes a Windows environment - there is a world outside Windows !!!

add to del.icio.us Add to Blinkslist add to furlDigg it add to ma.gnolia Stumble It add to simpy seed the vine TailRank


Developing with Gen as a framework

August 13, 2007

In a previous post, I expressed a belief that Gen should have a published API for other (non-CA) developers to produce an alternative set of generators so that the tool’s marketplace could be expanded.

Here, I’m focussing on the front end of things - i.e. the developer IDE.

Is there a place for an alternative IDE for the Gen tool ?

Certainly, portions of the toolset leave a lot to be desired, and some parts of the tool have already been re-engineered (the diagram trace facility as an Eclipse Rich Application), and alternative reporting tools and support tools like GuardIEN have matured.

BUT the user interface to the core development tool still leave me a little cold.

With the new releases of Gen, the advance into Eclipse has gathered momentum, but is that enough ?

There seem to be two opposing camps here - the Microsoft camp and the Eclipse camp. For me, used to a Microsoft-centric approach, it would seem a good move to me to adopt some of the Microsoft ideas, however there are those who favour the Eclipse approach would say that using Eclipse as a grounding for a development environment is the way forward.

But is there space for a third approach, a hybrid of them all ?

If CA were to keep control of the fantastic encyclopedia API and allow other developers to develop DEVELOPMENT interfaces to the tool, like they have done with other aspects, then 3rd parties could take the best of both worlds, and leave the worst of those worlds alone, and develop a dynamic, easy-to-use, productive, Gen-centric interface, with which to work.

 Maybe Gen could evolve into something which isn’t a tool, but a framework into which others’ tools slot in to produce the most effective working environment for the developer.

Those tools could include 3rd party interfaces, reporting tools, generators, and oher supporting items, all seamlessly integrating into a single framework, so that the developer/analyst/designer isn’t aware that they are using different vendors’ products.

add to del.icio.us Add to Blinkslist add to furl Digg it add to ma.gnolia Stumble It! add to simpy seed the vine TailRank


EDGEucate 2007 - Advance Programme available - download now !

August 10, 2007

What will you do at EDGEucate 2007 ?

  • Participate in more than 20 technical session and several Vendor Presentations that will keep you on top of the industry’s latest trends.
  • Discuss specific software needs with an impressive panel of company representatives at the EDGEucate Trade Show.
  • Network with your peers during the event and at the Tuesday Night Reception, sharing experiences and tips with fellow CA Gen users
  • Sign up to become an EDGE volunteer and foster the growth of this unique association!

See the CA-Gen’s most respected users, customers and CA partner vendors in a fantastic setting - Southfork Ranch made famous by JR and the rest of the “Dallas” cast in the 1980’s !

Download the advance program and registration details -

Click here for Hotel details

Don’t forget

EDGE EMEA 2007 in Ditton Park, London

September 30th - October 2nd 2007