More Reader Feedback On Control Your Code, And HMC Stay Of Execution

February 6, 2012 Hey, TPM

Don’t forget to mention COBOL programmers on the iSeries still writing green screens!

I always hear about the RPGers, but what about us COBOLers? Are there more out there? If you hear of anyone looking for a COBOL programmer, I’ve got 33 years experience in IT. I have about six years of experience writing RPG in the 1990s, too, but prefer COBOL. Been on the iSeries (AS/400) since it was introduced back in the 1980s. Glad to hear green screen is still alive and well elsewhere!

–Mary

Ah, Mary. I shoulda known better. Mea culpa.

–TPM

I always read your well-informed and well-written articles with interest and have done so since you first started writing them.

I am a generation older than you and joined IBM UK as a trainee -System/360 salesman in 1965. When I had finished my training, my market was investment banks and stockbrokers (I had worked at both beforehand). I had a very successful time as a salesman, selling major systems which I designed for my prospects. They were very advanced for their time, all data entry being on terminals rather than the punched cards that were normal at that time. One of the systems even included remote graph plotters, not supported by IBM then. I always enjoyed the technical side and reckoned that I knew at least as much as the IBM systems engineers who worked for me.

I later ran a computer service bureau where we developed an investment package (on /360). I believe that my team of developers was as good as any other, but I learnt just how slow and expensive system development was. I also learnt how difficult that it was to change systems to handle new requirements that had not been anticipated originally.

I concluded a long time ago that programming was the most inefficient process ever devised by mankind in any field of human endeavor. Fine for masochists (like me!), but not productive.

It occurred to me at some point that, unlike computers, the human brain coped, more or less, with whatever new concepts were thrown at it and that it did not have to be shut down, reprogrammed, tested, and restarted every time it learnt something new. If it did, we would know nothing and have no intelligence.

I eventually had the opportunity to try and build a system in which the database structure is defined in the database itself rather than in programs, and most functionality of any application is also defined in the same database, which also stores all user data. This means that new database structures and new applications or application changes can be created without any need to write any programs (about 95 percent of the time). No program code is generated. I had not been on a computer course since 1965 (and still have not been). Therefore I had an open mind on how to evolve a solution. I had no knowledge of AI and at first did not realize that my system was based on AI principles.

I have had it working in daily productive use since 1983. I built the first version on a System /38. I proved some years ago to IBM at several of their labs that what they estimated as a 10-month development task to build a simple hospital system could be done in two hours. They investigated it very thoroughly and eventually signed an agreement to remarket it round the world, but, for reasons that I was never able to discover, they did not announce it in any country so nothing happened. I doubt that there is anyone left in IBM who will remember this, but I have a copy of the license agreement!

This was very disappointing and I have had no commercial success. I have however developed the system further and it is much more advanced now. For instance, it generates HTML and JavaScript on the fly so all applications work immediately over the Internet and it is now a high availability system.

It is possible to build a business model that is accessible to all and so could be called open source. This prevents developers from spending so much of their time re-inventing the wheel. Multiple authorized users can access and add to this, allowing developers to use those parts of the business model that are suitable for their own needs and add their own modifications, without changing the base business model. There can be both public and private (i.e., used based) versions of all database and application definitions and these can be used together.

My product is called ERROS. There is so much more to it than I have described above but I think that it can solve the problems that you describe in your article. ERROS is extremely simple. There are just 10 DB2/400 physical files, all with identical layouts, one display file, one printer file and no logical files. The database can model extremely complex relational, hierarchical, and network structures in one integrated database.

The ERROS kernel programs are just 14.4 MB, and these, and the files, are not changed when new applications are created. There are also exit programs (only three at present) and these may need minor changes from time to time, but not often. You can create many new very complex applications without any program changes and without creating any new AS/400 objects–I still call it that!

There is so much more to it than I have described above, which I am happy to explain. I believe that adoption and expansion of my open source business model would be a real step forward for Power Systems-IBM i users.

I have not yet solved all the problems of the world (I am working on that!), but the computer industry, and more importantly our users, really needs a new way forward. I believe that ERROS can provide that.

It really does work.

I hope to hear from you.

Best wishes,

Rob Dixon, Erros Plc

OK, Rob. I am game. I will contact you and you can walk me through it. Readers, stay tuned.

–TPM

I think it may be worth mentioning that the C line of power 7 hardware was only supported by HMC and not supported by SDMC. That could be another reason the HMC survives another life. . . .

–Name Unknown

Hi Tim,

For the benefit of those folks who still want a HMC it’s probably worth pointing out that they CAN be virtualized with a little tweaking. Here’s at least one way to do it (which you will find at http://aixwiki.org/wiki/Installing_HMC_in_VirtualBox_or_VMWare) and I have an HMC on my laptop for support and training purposes using this method and VirtualBox (since it’s free). This means you can use any hardware you like and give it as many cores and as much RAM as you need to get it running well but typically without paying the premium price tag for the tin it runs on.

The SDMC, of course, can also be virtualized and IBM sells it that way as an option–it’s even a little cheaper I believe