On 2 Feb 99, at 8:49, Irena Thomas wrote:
> After that I downloaded from www.suttondesigns.com their freeware,
> Net2000 Software Scanner (n2000tr.exe) which was to detect which
> file could cause a problem next year.
I visited that web page and read their material:
In its free scanner mode, the program can legitimately flag any program
module that, somewhere in its code, obtains a current system date that
only provides two digits of year data. As you've seen, this includes
virtually every .COM, .EXE and .DLL on your machine. It is highly
unlikely that it can also detect user-input date fields that provide
only two characters for the year.
ONE way to "fix" the Y2K bug is to map the years 0-99 onto some other
range than 1900-1999. 1950-2049 or 1980-2079 are popular choices.
This is what they're referring to as "windowing".
In its paid-for form, the software patches code -- much the way a
virus does... -- to invoke this kind of "correction". NOTE THAT IT HAS
NO WAY OF KNOWING IF THE SOFTWARE *ALREADY* INCLUDES CODE TO PERFORM
SUCH A CORRECTION.
Note also that this only *postpones* the problem, rather than
eliminate it. [The section of their page titled "Not a Short-term
fix!" concedes that it's only good for 60 years.]
There are at least three forms of Y2K bug:
1. Failure to handle rollover from 99-Dec-31 to 00-Jan-01 correctly.
Windowing may help, but so will a reboot after the rollover.
2. Treating 2-digit dates less than 50, 60 or 80 as 19xx rather than
20xx. Windowing is an adequate solution, but their add-on windowing
*may* conflict with date-windowing code already present, and only their
solution only applies to dates retrieved internally in a few ways, and
probably not to dates input by users.
3. Failing to properly handle differences in date (ages) or 100 or
more years. Social Security and pensions clearly need to do this, and
genealogy or history programs might; date-windowing is not a useful
technique in this case.
The scanner program may fix problems in categories 1 and some in
category 2, but doesn't touch category 3. It cannot (very hard
theoretical problem) detect whether a category 1 or 2 problem is
already fixed, and risks adverse interaction with any such fix. And
the description of its mechanism is way "dumbed down", suggesting that
their target market is users not sophisticated enough to understand
their approach and its limitations.
The skills required to implement this approach are suspiciously
similar to those needed to construct a computer virus. It occurs to me
that a free downloadable Y2K checker program would make an ideal trojan
vector for virus delivery. [NOTE: I have no personal knowledge that
this is indeed the case; I just don't like the odds.]
Based on the description on their web page, I wouldn't touch this
product, or recommend it to others.
David G
Curious about the people moderating your
messages? Visit our staff web site:
http://nospin.com/pc/staff.html
|
