The Tortoise and the Hare
When I was in Boston a few weeks ago at the TPF Users Group Conference, I had the opportunity to spend some very productive time with Bob Cohen who is the Director of TPF Development in Danbury. After touching base on several related topics, the conversation became focused on a single issue - Availability vs. Reliability.
At various times in our careers, we've all found ourselves in a situation where our ability to continue with our own work is either hampered or inhibited by some other person or persons who haven't yet provided their "piece of the puzzle". As an example, consider the case of trying to test your programs when the driver and editor segments haven't been written yet. The point I'm trying to make is that we're all aware of the many dependencies that exist between programs and programmers alike. Very often, the programmer can put a "quick-and-dirty" together to allow the work to continue. In other cases, the programmer may have the option of taking the required code "as is", and make it work rather than sit around idle waiting for the final product. In both of these scenarios, the developer has to decide between the availability of the code and the reliability of the code.
As a TPF Systems programmer, the matter of availability versus reliability is more than a technical issue. In most cases, the TPF systems in use today are the bread and butter of the company running those systems. The business needs of an airline, or hotel, or bank very often dictate the technical requirements of the operating system. But let's get specific. It's no secret that IBM has been diligently working for some time now on the Next Big Release of TPF. While I can't speak about any details of NBR, many of us recognize the benefits and enhanced functionality that NBR will provide. While no formal date has been announced by IBM for the general availability of the product, I think it's safe to say that we're at least a year away from getting the release from Danbury.
Certainly, there are shops today that are facing some of the constraints and limitations with the current 3.1 product, and there are probably other installations that would take advantage of the many enhancements that NBR will provide if it were available today. Accepting the fact that Danbury is still in a development mode with NBR, are there really some customers out there that would take the code today, if it were offered? In very simple terms; Do I want the code now, or do I want the code to work? Now let me qualify the question to be fair to everyone. I don't mean to imply that by taking a pre-release of the code, none of it would work. What I do mean, is that any pre-release of the code would be initiated with the following understanding:
To offset these negatives, one also needs to consider the advantages of a pre-release:
Of course there are the factors of time and money which must be invested in an undertaking of this magnitude. The implementation of any new release of TPF is not a trivial task. The allocation of human resource, the many months of testing, debugging, retro-fitting of in-house modifications, and potential changes to existing applications can very quickly become an expensive endeavor.
I don't have the answer. Each shop must weigh the value of the opportunity and decide what's best for them. Is it better to assume the role of the Hare, and take the availability approach, or do you emulate the patience and determination of the Tortise, and set your sites on reliability? I would really like to hear your thoughts on the matter.
If you have an opinion on which approach is the best, I would appreciate it if you would put it down on paper, and send it to me in care of this newsletter. I'll not only print your responses in the next issue, but also pass your letters on to Bob Cohen in Danbury. Thanks!