Tuesday, November 29, 2011

this is an undefined entry

Standardization is indispensable. Through independent standards, various products can work well, regardless of market competition. On the other hand, successful standards have appropriate limits. Too much constraint impairs vital product flexibility1. To balance these concerns while remaining unambiguous, a standard might literally delineate topics which are called "undefined". It identifies and then ignores. Hence, a well-known technological gaffe is to assume too much about what's undefined by the standard.

I appreciate such indistinct confessions. In fact, one of my cynical unfulfilled wishes is to exercise that option in other contexts.
  • I'll finish the project by the agreed deadline. On the condition that everyone else finishes the preceding tasks promptly. Else the finish date is undefined.
  • I'll contribute a proportional share to the joint cost of supper. On the condition that I stopped by an ATM recently. Else I should warn you that the amount that I will find in my wallet is undefined.
  • I'll advocate the replacement of the software package. On the condition that the result is an unqualified improvement. Else my original level of support for the replacement is undefined.
  • I'll follow the specifications. On the condition that the specifications are exact matches for what I was already planning. Else my fidelity to the specifications is undefined.
  • I'll write a program that handles the input correctly. On the condition that all the input is correctly entered. Else the program's actions are undefined.
  • I'll stay alert at all times in long meetings. On the condition that I slept soundly the previous night. Else my attentiveness is undefined.
  • I'll continue adding more examples to the list. On the condition that I don't start to feel bored. Else the number of more examples is undefined.


1The product can exploit the flexibility in myriad ways. It could do nothing or fail. Or it could provide additional benefit. It could violate unstated expectations. Or it could guarantee a result which the standard doesn't. It could optimize for greater efficiency by performing fewer checks. Or it could optimize for greater infallibility by performing innumerable checks.

No comments:

Post a Comment