Oops! Y2K redux, the 2^30 problem, due 10 January

[Winfield Hill]

Some programs evaluate time by counting seconds since
1970. And some programs use a 30-bit number to count
these seconds (i.e. 32 bits less 2 bits for flags).

On 10 January next, a few weeks off, these programs
will experience serious Y2K problems. As an example
see here, http://news.com.com/2100-7355-5129875.html

How about your software, will any of it be affected?

Thanks,
- Win

whill_at_picovolt-dot-com

 

[Winfield Hill]

Winfield Hill wrote...

Some programs evaluate time by counting seconds since
1970. And some programs use a 30-bit number to count
these seconds (i.e. 32 bits less 2 bits for flags).

On 10 January next, a few weeks off, these programs
will experience serious Y2K problems. As an example
see here, http://news.com.com/2100-7355-5129875.html

And see http://www.ptc.com/go/timeout/index.htm

How about your software, will any of it be affected?

Thanks,
- Win

whill_at_picovolt-dot-com

 

[Jim Thompson]

Some programs evaluate time by counting seconds since
1970. And some programs use a 30-bit number to count
these seconds (i.e. 32 bits less 2 bits for flags).

On 10 January next, a few weeks off, these programs
will experience serious Y2K problems. As an example
see here, http://news.com.com/2100-7355-5129875.html

How about your software, will any of it be affected?

Thanks,
- Win

whill_at_picovolt-dot-com

So *who* is PTC?

...Jim Thompson

 

[Ken Taylor]

So *who* is PTC?

...Jim Thompson
--

Clearly a software company looking for a new income stream. Man, that whole
Y2K thing was such a wank......

Ken

 

[Mike]

Some programs evaluate time by counting seconds since
1970. And some programs use a 30-bit number to count
these seconds (i.e. 32 bits less 2 bits for flags).

On 10 January next, a few weeks off, these programs
will experience serious Y2K problems. As an example
see here, http://news.com.com/2100-7355-5129875.html

How about your software, will any of it be affected?

Wow, I can see why you'd be concerned: "PTC engineers discovered that the
flaw was widespread, affecting most of the company's 35,000 customers
worldwide who use PTC products such as Pro/Engineer, Pro/Intralink and Win
Hill."

Oh, sorry. That was "Windchill."

Never mind.

-- Mike --

 

[Mike]

Some programs evaluate time by counting seconds since
1970. And some programs use a 30-bit number to count
these seconds (i.e. 32 bits less 2 bits for flags).

On 10 January next, a few weeks off, these programs
will experience serious Y2K problems. As an example
see here, http://news.com.com/2100-7355-5129875.html

From the article: "Gavaghan said the patches will reset the infinity value
to 4 billion seconds, buying current PTC products another few decades of
life. Subsequent releases will eliminate date dependency, he said."

Eliminate date dependency? How do they do that?

-- Mike --

 

[Tom Del Rosso]

In Mike typed:

Wow, I can see why you'd be concerned: "PTC engineers discovered that
the flaw was widespread, affecting most of the company's 35,000
customers worldwide who use PTC products such as Pro/Engineer,
Pro/Intralink and Win Hill."

Oh, sorry. That was "Windchill."

Never mind.

It's just Win Hill with a DC component.

 

[Rather Play Pinball]

We only buy *quality* software from Microsoft, so we won't face this
problem.

<insert cough, hack, throw-up sounds here>

;-)

- - -

 

[Robert Reimiller]

On 21 Dec 2003 18:07:45 -0800, Winfield Hill
<[email protected]> wrote:

So *who* is PTC?

It thought it was Positive Temperature Coefficient

 

[Tim Hubberstey]

From the article: "Gavaghan said the patches will reset the infinity value
to 4 billion seconds, buying current PTC products another few decades of
life. Subsequent releases will eliminate date dependency, he said."

Eliminate date dependency? How do they do that?

Newer Unix releases do it by using a 64-bit value for the time. While it
doesn't technically eliminate the issue of rollover, it does move it out
some 584 billion years, well past the predicted end of the universe. I
doubt that there will be any repercussions. ;-)

By the way, the next "critical" dates are 2038 (for systems using a
32-bit signed number for seconds since 1970) and 2106 (for systems using
a 32-bit unsigned). I wonder if these dates will generate the same
hysteria as 2000 did? Hope I'm still around to see the fun.

 

[Ken Taylor]

In Mike typed:

It's just Win Hill with a DC component.

Bravo!

Ken

 

[Spehro Pefhany]

Bravo!

Ken

A splendid example of tmesis.

Best regards,
Spehro Pefhany

 

[Winfield Hill]

Ken Taylor wrote...

Tom Del Rosso wrote ...

Bravo!

ROFLOL!

Windchill with wind-speed = 0 ... ?

Thanks,
- Win

whill_at_picovolt-dot-com

 

[Robert Baer]

Some programs evaluate time by counting seconds since
1970. And some programs use a 30-bit number to count
these seconds (i.e. 32 bits less 2 bits for flags).

On 10 January next, a few weeks off, these programs
will experience serious Y2K problems. As an example
see here, http://news.com.com/2100-7355-5129875.html

How about your software, will any of it be affected?

Thanks,
- Win

whill_at_picovolt-dot-com

Err....The original MSDOS for the original IBM PC/XT (back in the
1980s) coded file time based from Jan 1, 1980 and the BIOS also worked
that way.
Everybody *slavishly* copied that, making it a defact standard.
I have never seen any ?program? that "pre-dated" that to start
counting from 1970!
So, methinks you need to correct your statement, *or* disclose the one
stupid program (and author) that screwed up.

However, make the statement act like a school teacher, and give it
"tenure" (ie: give ten more years).

 

[Robert Baer]

From the article: "Gavaghan said the patches will reset the infinity value
to 4 billion seconds, buying current PTC products another few decades of
life. Subsequent releases will eliminate date dependency, he said."

Eliminate date dependency? How do they do that?

-- Mike --

Obviously, they will no longer go out on dates.....

 

[Joe Legris]

Err....The original MSDOS for the original IBM PC/XT (back in the
1980s) coded file time based from Jan 1, 1980 and the BIOS also worked
that way.
Everybody *slavishly* copied that, making it a defact standard.
I have never seen any ?program? that "pre-dated" that to start
counting from 1970!
So, methinks you need to correct your statement, *or* disclose the one
stupid program (and author) that screwed up.

However, make the statement act like a school teacher, and give it
"tenure" (ie: give ten more years).

The article mentions that UNIX systems use 1970 as the year zero.

Aside from this issue, isn't it rather amazing how the authors of UNIX
invented an O.S. that anticipated the needs of networked personal
computers more than a decade before they existed and even now 35 years
later?

 

[Ian Stirling]

Robert Baer wrote:

The article mentions that UNIX systems use 1970 as the year zero.

Aside from this issue, isn't it rather amazing how the authors of UNIX
invented an O.S. that anticipated the needs of networked personal
computers more than a decade before they existed and even now 35 years
later?

This isn't quite true.
The original unix had no networking support.
I'm not completely sure when TCP/IP support went in.
It was by no means certain that there would be an internet as widespread
as it is now.

Early on, AOL and various others were heavily pushing their own networks,
and were only forced by the growing internet to connect to it.

If the internet had been delayed in its growth by a couple of years, to
allow AOL et al to consolidate (as may have happeed if Gore had not
implemented certain changes (but certainly not invented)) then things might
be a whole lot different.

AOL might have gained critical mass in rolling out its network and getting
lots of subscribers before competitors could get in the door.

Most people may have ended up connected to AOL, with only a few limited
gateways to the wider net.
And given lack of investment, the net would be a lot smaller.

 

[John Woodgate]

I read in sci.electronics.design that Winfield Hill
about 'Oops! Y2K redux, the 2^30 problem, due 10 January', on Mon, 22
Dec 2003:

ROFLOL!

Windchill with wind-speed = 0 ... ?

Wind speed = constant?

 

[John Woodgate]

I read in sci.electronics.design that Robert Baer

Everybody *slavishly* copied that, making it a defact standard.

Since we're into portmanteau words, that should be 'de fecto standard'.