It must be within that range somewhere. Keep looking. posted by scottreynen at 7:12 AM on December 17, 2006 [1 favorite]
theora55 pointed the error out previously; apparently it's been an issue for a while. posted by Mitheral at 7:29 AM on December 17, 2006
matt, just subtract .0000000000000098 from everything. posted by quonsar at 7:35 AM on December 17, 2006
anastasiav has the same error and thinks it's a problem with the user near me code. posted by Mitheral at 7:36 AM on December 17, 2006
And you guys live near each other, actually. posted by jessamyn at 8:00 AM on December 17, 2006
yeah, there's some huge math function to determine users near me and for some reason, a few people in the northeast end up making it divide by zero or something.
I'll see if I can trap that error and make it go away for now. posted by mathowie at 8:29 AM on December 17, 2006
a few people in the northeast end up making it divide by zero
damn yankees. posted by sonofslim at 8:45 AM on December 17, 2006
Matt, do you recognize that there is a difference between 1.0000000000000002 dollars and 1.0000000000000002 cents? posted by arcticwoman at 10:05 AM on December 17, 2006
yes, but he does not recognize that there's a difference between exceeding an allowable range by .0000000000000002 and dividing by zero. posted by quonsar at 10:36 AM on December 17, 2006
yes, but he does not recognize that there's a difference between exceeding an allowable range by .0000000000000002 and dividing by zero.
He said divide by zero or something posted by delmoi at 11:16 AM on December 17, 2006
This is just your garden variety floating point confusion. Notice that there are 15 zeros, so the .2 is in the 16th place. Also remember that a standard IEEE 754 double can only represent approximately 15 digits of precision. So this value is really 1.0, and the 2 on the end is a rounding error that came up because of flawed algorithms. The flaw is doing things like "if (a - b) = 0" when dealing with floating point values. You have to change that to "if (a - b) < (smallest representable number in the precision used, epsilon).br>
What Every Computer Scientist Should Know About Floating-Point Arithmetic
It's amazing how little people actually know about how floating point math works. posted by Rhomboid at 11:43 AM on December 17, 2006 [1 favorite]
MetaTalk: divide by something posted by Ceiling Cat at 12:36 PM on December 17, 2006
This is just your garden variety floating point confusion.
GEEEEEEEEEEEK. posted by chrismear at 1:58 PM on December 17, 2006
Almost there! Keep trying! posted by armoured-ant at 2:04 PM on December 17, 2006
Yep, this bug has been around for a while. Presumably it only happens to a few accounts, otherwise I suspect there would be more complaints. posted by RichardP at 2:52 PM on December 17, 2006
Subscribe
posted by Civil_Disobedient at 6:52 AM on December 17, 2006