5 ms·

the advantage you sight is exactly problem why the international foot and U.S. survey foot exist and the issue mentioned in article happens. it may be easy for

the advantage you sight is exactly problem why the international foot and U.S. survey foot exist and the issue mentioned in article happens. it may be easy for us to think in fractions with repeating decimals, or irrational numbers ,it is super hard for computers which do Floating Point arithmetic, fractional arithmetic is not common from assembly, higher languages, or libraries we typically use to develop apps. This is why engineering is always in SI, and errors happen in the interface between people refusing to change, Given that most computing well happens in computers, It is lot more convenient for us change to a system computers understand, rather than keep a system we understand

Quarrel 6y ago
> engineering is always in SI Sadly several significant NASA failures would disagree.
- ghaff 6y ago
  From quite a while ago. (The last failure I can think of related to units was over 20 years ago which means the design mistake was probably more like 25 years ago.)
a1369209993 6y ago
> a system computers understand Um, what? Computers understand base 2 (or 16 for compactness), not base 10. A centimeter is not 0.00999999977648258209228515625 meters.
- moring 6y ago
  Computers can be made to understand base 10 though, if needed. We do so routinely when amounts of money are involved so a cent is a cent, even though 1/100 is nonterminating as a binary fraction.
  - a1369209993 6y ago
    By that logic they can be made to understand base twelve (et al) just as well. And 0x64 cents is perfectly reasonable amount of money, but that's because it's a integer, not a fraction.
  - Gibbon1 6y ago
    Brings up someone I read from an old crusty programmer, binary floating point is the wrong tool for most things. It's fine for scientific simulation work. But for everything else it causes no end of problems.
    - a1369209993 6y ago
      Floating point, binary or otherwise, is the wrong tool for many things. (Although (not-necessarily-scientific) simulation work is a very broad class of things.) If you're doing anything related to accounting, there is no excuse not to be using integers.
- manquer 6y ago
  There are problems of fractional arithmetic in any base including binary. binary fractions are not base10 fractions and fraught with rounding errors, one more reason to not to use fractional arithmetic . The U.S. foot mentioned in the article is defined as 1200/3937 of a meter which is 0.304800609601... depending on the precision you want the number will keep changing. In Imperial you will find a lot of unit conversions like this which will need fractional arithmetic in base 10 like a foot is 1/3 yard etc. You will not find that in SI units as the base(10) and unit conversions (10) are the same making the definitions precise and avoid fractions between unit conversions

Quarrel 6y ago

> engineering is always in SI Sadly several significant NASA failures would disagree.

ghaff 6y ago
From quite a while ago. (The last failure I can think of related to units was over 20 years ago which means the design mistake was probably more like 25 years ago.)

a1369209993 6y ago

> a system computers understand Um, what? Computers understand base 2 (or 16 for compactness), not base 10. A centimeter is not 0.00999999977648258209228515625 meters.

moring 6y ago
Computers can be made to understand base 10 though, if needed. We do so routinely when amounts of money are involved so a cent is a cent, even though 1/100 is nonterminating as a binary fraction.
- a1369209993 6y ago
  By that logic they can be made to understand base twelve (et al) just as well. And 0x64 cents is perfectly reasonable amount of money, but that's because it's a integer, not a fraction.
- Gibbon1 6y ago
  Brings up someone I read from an old crusty programmer, binary floating point is the wrong tool for most things. It's fine for scientific simulation work. But for everything else it causes no end of problems.
  - a1369209993 6y ago
    Floating point, binary or otherwise, is the wrong tool for many things. (Although (not-necessarily-scientific) simulation work is a very broad class of things.) If you're doing anything related to accounting, there is no excuse not to be using integers.
manquer 6y ago
There are problems of fractional arithmetic in any base including binary. binary fractions are not base10 fractions and fraught with rounding errors, one more reason to not to use fractional arithmetic . The U.S. foot mentioned in the article is defined as 1200/3937 of a meter which is 0.304800609601... depending on the precision you want the number will keep changing. In Imperial you will find a lot of unit conversions like this which will need fractional arithmetic in base 10 like a foot is 1/3 yard etc. You will not find that in SI units as the base(10) and unit conversions (10) are the same making the definitions precise and avoid fractions between unit conversions