| 1 |
On Sunday 30 March 2014 16:07:49 Rich Freeman wrote: |
| 2 |
> Proposal 1 |
| 3 |
> "Whenever practical Developers are encouraged to use SI units and base |
| 4 |
> 10 values (ie 1KB = 1000 bytes). They may use base 2 values when this |
| 5 |
> output is more likely to be useful to users (eg in memory hexdumps, |
| 6 |
> etc). Either way, unit prefixes defined in IEC 80000-13 (KB, KiB, |
| 7 |
> etc) must be used so that output is unambiguous. This does not |
| 8 |
> require maintainers to patch upstream code to change its behavior, but |
| 9 |
> they should be applied with code that originates in Gentoo." |
| 10 |
> |
| 11 |
> While I understand the resentment at the redefinition of prefixes that |
| 12 |
> have been in use for decades, the ambiguity of using SI prefixes with |
| 13 |
> non-SI definitions creates confusion and potentially error. I think |
| 14 |
> clarity should always be valued when the change is otherwise cosmetic. |
| 15 |
> |
| 16 |
> So, use MiB or MB as makes sense, but the latter should be the default |
| 17 |
> and should always mean 1000000 bytes. |
| 18 |
|
| 19 |
+1 |
| 20 |
|
| 21 |
Whatever terminology we use, it had better meet the requirement that one |
| 22 |
kilobyte per kilometer equals one byte per meter. And the definition of the |
| 23 |
kilometer isn't going to change. |
Archives