Reading moodily one of yesterday's posts "Contest - when did you first see windowing", I am in total disagreement with two of the technical statements made in this thread. So much so that I am going to go back on my long-standing decision to mention my Web site www.crosslink.net/~erington and each of its papers only one time on this forum.

ng, the poster, says "I can see encapsulation being patentable - also hard to use". Well, I don't think encapsulation is any more patentable than windowing, but it's the judgement that it is hard to use that I have a quarrel with. One of the papers on my site, "A Hell of a Fine Stopgap", goes into great detail, step by step, how easy this approach is.

A responder, Jim, says regarding windowing, that "if you change the pivot year from 00 to, say, 30, then you have a Y2030 problem just as big as Y2K". Another paper on my site, "Only a Temporary Repair?", has as its point that this argument (applying to both windowing and encapsulation) is fallacious.

Having said this, I am in total agreement with every opinion that I have read on this forum that a patent on windowing is ridiculous. Let me illustrate with my own personal history.

In late 1996 and then in very early 1997, I came up with the ideas for windowing and encapsulation all by myself, with all the details as to how I would implement these two approaches. I didn't even know what these two approaches were called then. I do not make this statement to parade my own brilliance, far from it. I make this statement because these two approaches are anything but Einsteinian, they are both pretty damn obvious, really. When I got these two ideas, with hardly any analytical effort, I figured that many other people clearly must have also had these ideas before me. On another thread yesterday concerning the windowing patent ("Firms May Face Fees For Using Patented Y2K Fix") I'm Here I'm There states "From the very first day I started working on this (around 1994), there were already three accepted ways of fixing the problem: Windowing, Encapsulation, and Expansion". Well, there you are.

-- Peter Errington (petere@ricochet.net), November 03, 1999

Answers

Peter,

The reason I said encapsulation could be patentable was the first time I saw the technique described, it was described in detail by a person who claimed to have invented it, with the term encapsulation attached. This was around 1992. Windowing, on the other hand, was probably used by the insurance industry on wired accounting machines in the late 50s. Heck, windowing has been used by humans as mental logic since year was first represented by less than 4 digits.

Most of the maintenance projects I have been associated with (many) have defaulted to windowing. Primarily because it lends itself beter to incremental development on large complex systems better than the other two. The jury is still out on if it introduces more bugs, it may.

Speaking of expansion, early on it was the preferred solution, then people started looking at the impact on synchronization of system changes and data conversions and interfaces...

Have you or anyone seen widespread use of encapsulation on large complex systems?

Thanks for your insight.

ng

-- ng (cantprovidewemail@none.com), November 03, 1999.

Peter,

I have not read your papers and am by no means an "expert" on windowing, most of my work being expansion. When I wrote that comment and sent it I immediately realized it was very much in error: If the pivot year was 30 and the organization was approaching 2030, say 2025, I'm assuming they could switch the pivot year to 40 or 50 and keep continuing. Not that it would be a single, one minute quick fix - I'm sure there would still be some analysis and assessment work to locate all the windowing pointers and references to be sure they were all gotten. But this "Y2030" problem would be much less in scope than Y2K - not even mentioning the absence (if that's even correct) of the embedded problem.

With windowing, we would eventually be approaching the 21st & 22nd century "boundary", so the actual centuries implied by the pivot would need to be changed (from 19 & 20 to 20 & 21). I also know that one pivot year cannot handle all date fields. Some are very aged data (birthdates, historical records, etc.), some are near current (most customer records, invoices, payments, etc.), some are future oriented (mortgages, loans, contracts, etc.). There is also the point of data transfers, and what a two digit year means on your system may not necessarily mean the same on another system which shares the same data. All of these points are brought up in excellent detail on the Dale Way IEEE paper.

My point was that windowing sucks for the long term. Sure its quick and cheap right now, but isn't that how we got into Y2K in the first place?? Windowing only adds (unnecessarily) to the complexity of the IT environment. Expansion eliminates all ambiguity. The only place for windowing in an IT environment is in data entry. Everything else should be mandated to be expansion once we get past the Y2K event horizon and we have the time to do such.

-- Jim (x@x.x), November 03, 1999.

This is a test to link directly to a post in the original thread. It is my last comment in the original thread on windowing.

-- Jim (x@x.x), November 03, 1999.

This is a test to link directly to a post in the original thread. It is my last comment in the original thread on windowing. (Scroll up due to first failed test).

-- Jim (x@x.x), November 03, 1999.

ng, in response to your question, do I know of a case where encapsulation has been used? The answer is no; the organizations that I am aware of are all doing windowing. I would be very interested in hearing from anyone with knowledge of cases where encapsulation has been used.

-- Peter Errington (petere@ricochet.net), November 03, 1999.