[Previous Months][Date Index][Thread Index][Join - Register][Login]
[Message Prev][Message Next][Thread Prev][Thread Next]

Re: [IP] year 2000 problem (long)

Pardon my being a bit thin-skinned here, but being a software engineer who
has done some Y2K work I find you remark about "software engineers trying
to make a pile of money" out of line. 

Foremost, it's not the SEs who are blowing this "way out of proportion".
Qualified SEs know what is the scope of the issue and don't overstate
OR understate the problem. The overplay comes either from media ninnys
who have to sensationalize everything they don't understand or from 
management gerbals who want to make sure it "gets the right attention"
and they don't get blamed for it.

That being said, there is cause for significant concern over the Year
2000 problem (or in buzz-word terms, "Y2K"). It is a problem of comparison.
Much of the software in various industries stores dates as two digits.
The last two digits of the four digit year(i.e. 98 for 1998). If you
just do math with those two digits when the year 2000 rolls around, you
get very unpredictible results (00 - 99 = ?). 

If your business depends on such calculations, you should currently be
well on your way to replacing two digit data and calculations in your
software or well on your way to having bankruptcy paperwork prepared.

On the other hand, if your a gigantic government bureaucracy, say the
U.S. Social Security Administration, you probably have been too busy 
planning your 2000 New Years party to worry about it. Think I'm being
facetious? Try this on for size. Some agencies of the U. S. Federal
government are predicting that they will finish their Y2K conversions
in 1998?, 1999?, 2000? Nope, try 2010. If you were planning on retiring
in 2003, you may not get your first social security check until 2012.

Now as far as real-time devices such as pacemakers and insulin pumps
are concerned, there isn't a whole lot of reason to panic. For one
thing, they don't tend to do many comparisons across dates. If you 
don't worry about dates, you don't have a Y2K problem.

The cyclic events generated by real-time devices, like basal insulin 
injections, are almost invaribly tied to the internal clock of the
microprocessor running the device. The time it takes for one "tick"
or cycle of that internal clock to occur is a fixed value. By knowing
that time, and counting how many cycles have occurred since the last
event, the computer can calculate when the next event is to occur.

The main point is this - the pump doesn't use human-readable
dates to calculate when you get insulin. If it does, I'm going back
to using a syringe. 

Both your meter and you pump generate dates for their displays, but
that's easy and doesn't involve comparisons. If the meter can't add
one to 1999 and get 2000, I definitely am not going to trust the
bg readings I get out of it.

The moral of the story is "Don't Worry, Be Happy"! Unless your a
corporate executive, a high-ranking government official, or a
(money grubbing ;-{) software engineer, you can't do anything about
the problem. But you might want to lay in some extra pump supplies
since a number of government agencies and insurance companies will
decide on Jan 1, 2000 that you don't need any supplies because you
won't be born for 50 or 60 years.

> Date: Fri, 8 May 1998 00:43:10 -0400
> From: email @ redacted
> Subject: Re: [IP] year 2000 problem
> Hey lighten up John.  This year 2000 problem has been blown way out of
> proportion by software engineers trying to make a pile of money.  (And they
> are doing just that).  There's more than enough for us diabetics to worry
> about today than losing the memory of 100 glucose tests a year and a half
> from now!  If one were really concerned,  just take a printout on New Years
> Eve.  Duh. We already know there's no problem with Minimed.   I bet your
> mother worried about your getting razor blades in apples when you went
> trick or treating :-)
> As FDR said (was he diabetic?) , "We have nothing to fear but fear itself."
> - -wm
> <<<From: John Neale <email @ redacted>
> Subject: Re: [IP] Millenium bug (sorry, long)
> With the greatest respect, I find your head-in-the-sand ignorance about
> potential millenium bug problems alarming. There are loads of things
> that could go wrong on 1/1/2000 with any piece of equipment that holds
> the date. For computers, the rollover from 1999 to 2000 is often a
> unique moment in date handling routines. That's where bugs show up and
> programs crash. The cautious and sensible are assuming that any date
> sensitive equipment or software WILL fail, until they have tested it,
> and verified that it DOESN'T fail.
> I asked my question because, while I know that my Minimed507 doesn't
> hold the date, I've no idea about other pumps and models. Bg meters
> often do hold the date (my Dex does) so there's a potential problem
> there until you've checked it out. You may find you lose the record of
> the last 100 tests, or at worst the software in the meter could simply
> stop working, and the meter become useless.>>>>

Insulin-Pumpers website http://www.bizsystems.com/Diabetes/
For subscribe / unsubscribe information,
send the next two lines in a message
to the e-mail address: email @ redacted