What is causing the 2010 bugs?

Question

There are a lot of reports of systems failing to understand the year 2010 but I've no idea why. Current systems I look after are working fine as far as I could tell but I'd like to know what the actual problem is to search better.

Could anyone shed some light on it please?

Edit: http://www.rte.ie/business/2010/0105/bug.html - Information about it affecting credit cards in Germany

Oh dear, that's awfully vague. Do you have any links to share? — Tad Donaghe, Jan 05 '10 at 16:04
I heard something about interpreting `10` as hexadecimal number. — Gumbo, Jan 05 '10 at 16:04
Maybe consultants added these bugs back when they fixed Y2K, so they'd have a bunch of easy work to do 10 years later. — Kaleb Brasee, Jan 05 '10 at 16:05
maybe it doesn't fit this site, but an interesting question nontheless. — peterchen, Jan 05 '10 at 16:08
Smart move, by jumping from 2009 to 2016 we mis the destruction of the world in 2012. This programmer deserves the nobel peace price! — Toon Krijthe, Jan 07 '10 at 09:40

score 13 · Answer 1 · answered Jan 05 '10 at 16:12

13

One possible explanation is in the article below

http://www.theregister.co.uk/2010/01/05/symantec_y2k10_bug/

Reminds me of your recent article about cheap and dirty Y2K bug fixes where some unscrupulous programmers put in a simple if <10 = 20xx otherwise the date is 19xx

answered Jan 05 '10 at 16:12

Richard Ev

52,939
59
191
278

Windows, currently, has the same thing. Two-digit years are intrepreted as a date between 1930 and 2029. Which makes the code `if < 30 = 20xx otherwise the date is 19xx`. The fix is to smack people, willingly, using 2-digit years (including end-users) – Ian Boyd Jun 22 '10 at 16:06

score 13 · Accepted Answer · answered Jan 06 '10 at 15:16

Several protocols used in banking and telecommunications - including the SMS protocol - encode the year as BCD in a single byte.

From 2000-2009 one could easily make the mistake of interpreting the year as a standard binary number since the encoding would be the same:

Encoding  Binary-interpreted  BCD-interpreted
0x01      2001                2001
0x02      2002                2002
...
0x09      2009                2009
0x10      2016                2010
...

That is most probably the cause of the Windows Mobile bug.

score 11 · Answer 3 · answered Jan 05 '10 at 16:14

11

SpamAssassin had a rule to mark dates too far in the future as spam:

/20[1-9][0-9]/

The fix came a few days too late, but it's quite simple:

/20[2-9][0-9]/

See you again in ten years.

answered Jan 05 '10 at 16:14

Mark Byers

811,555
193
1,581
1,452

2

First I though that you must be kidding. But then I checked https://issues.apache.org/SpamAssassin/show_bug.cgi?id=6269. Doesn't seem too hard for me to compare two 4-digit integers with each other. Why do they use regular expressions for that? – Dirk Vollmar Jan 05 '10 at 16:23
14

Because everybody knows that, if you can't express it in a Regex, it's just not worth doing... ;) – GalacticCowboy Jan 05 '10 at 16:33
May be to avoid string to integer conversions? – Alex. S. Jan 05 '10 at 16:47
3

Why not just sample the system date, and base the "too far in the future" off of that? WTF? – Dean J Jan 06 '10 at 15:19

score 4 · Answer 4 · answered Jan 05 '10 at 16:07

4

I've got a system at work that uses a one digit year field. Yes. One digit. So the reason this system is failing is that "2000" is expressed the same as "2010".

answered Jan 05 '10 at 16:07

recursive

83,943
34
151
241

4

Is/was there any kind of rationale for that decision? – Paolo Jan 05 '10 at 16:10
I have no idea, and no real way of finding out. – recursive Jan 05 '10 at 16:13
11

"We'll never be using this software 10 year from now." – DA. Jan 05 '10 at 16:21
1

Some versions of the TRS-80 DOS (like TRSDOS 6) used one octal digit (three bits) to represent the year. That was one of the things pushing me to get off the TRS-80 Model 4 and onto a Macintosh. – David Thornley Jan 05 '10 at 16:22
The AT&T 6300 had the same problem: The internal clock could only represent years between 1984 and 1991. – dan04 Jun 23 '10 at 04:31

score 4 · Answer 5 · answered Jan 05 '10 at 16:11

4

The one I heard about was quick fixes people did for Y2K without thinking it through. So if xx < 10 then 20xx else 19xx.

answered Jan 05 '10 at 16:11

Rich Adams

26,096
4
39
62

1

Doesn't really exaplain the 16 – Martin Beckett Jan 05 '10 at 16:12
1

Where's 16 come from? Question just asks about systems failing to understand 2010. 0x10=16 could be one reason, sure. – Rich Adams Jan 05 '10 at 16:18
The 2010 to 2016 bug may not be the only bug here... This could very well explain the problem the German banks are having. – Austin Salonen Jan 05 '10 at 16:19

score 4 · Answer 6 · edited Jan 05 '10 at 16:25

4

It might be due to the young developers who started their careers after Y2K and are using 1 digit to represent the year.

edited Jan 05 '10 at 16:25

Richard Ev

52,939
59
191
278

answered Jan 05 '10 at 16:16

Larry Watanabe

10,126
9
43
46

2

Wouldn't this be more characteristic of an older developer who has more experience (a) minimizing these representations and (b) developing the systems where this is more likely to happen? – Austin Salonen Jan 05 '10 at 16:34
2

Or it might be the Y2K consultants who didn't make enough to retire :) – Larry Watanabe Jan 05 '10 at 17:14

Tj Kellie · Answer 7 · 2010-01-06T06:24:51.070

4

I took care of a little 2010 fail in a site last weekend, it was just the result of an oversight in coding though.

Someone thought it would be a good idea to set the value of a list item to the current dateTime.year.Now() when the list only contained items up to 2009.

ddlItem.findByText(DateTime.Now.Year.ToString())

edited Jan 06 '10 at 06:24

answered Jan 05 '10 at 16:19

Tj Kellie

6,336
2
31
40

score 2 · Answer 8 · answered Jan 05 '10 at 16:29

Here is a screen shot of the norton symantec endpoint protection

alt text http://img695.imageshack.us/img695/4500/152010112800am.jpg

Really nice that no one @ symantec informed their customers... Till the article was posted: http://www.theregister.co.uk/2010/01/05/symantec_y2k10_bug/

score 2 · Answer 9 · answered Jan 05 '10 at 16:59

2

It is that there is a bug in a component that splits the year in two parts. The second part is used in a comparison so that the digit 10 is not in base 10, it's in base 16 meaning that it's 0x10 = 16 (hex).

answered Jan 05 '10 at 16:59

Tower

98,741
129
357
507

score 2 · Answer 10 · answered Jan 07 '10 at 20:42

I used Google Code Search to find y2010 bugs in open source software. I looked for one particular pattern that would indicate a bug (use of "200%d" as a printf format string), and found several projects with that bug. Creative application of search patterns could probably turn up more different kinds of bugs.

What is causing the 2010 bugs?

10 Answers10

Linked