FC5 test1 relnotes for translation

[David Woodhouse]

On Tue, 2005-10-25 at 12:36 -0700, Karsten Wade wrote:
> Therefore, if you want to influence/convince FDP to use MiB instead of
> MB, please feel free to do so.  

OK :)

> Whatever consensus we reach becomes our standard, and we can start a
> Wiki page on the subject.  Again, we would be doing the right thing
> and trying to lead the way with standards. Makes sense on the face of
> it.

The standard in this case is the second edition of IEC 60027-2,
discussed at http://physics.nist.gov/cuu/Units/binary.html

Using the SI decimal prefixes for binary multiples is, strictly
speaking, incorrect. That just isn't what they mean.

Obviously in some cases it doesn't really matter much or it's obvious
from the context -- we'd never actually _mean_ '128MB', for example, in
the context of the amount of physical RAM available in the system. It
would always be 128MiB. But that still doesn't give any reason for
choosing to continue to get it wrong, even in those cases.

There's a reason to get it _right_ though, aside from the simple fact
that we should be striving to be pedantically correct just to quiet the
voices in our heads -- or is that just me?. By being strict about it, we
can ensure that our meaning is perfectly understood in those (admittedly
relatively infrequent) cases where it _does_ matter.

When you see a binary 'Mi' used, you know precisely what it means --
it's a power-of-two multiple. On the other hand, when you see the SI
decimal prefixes used, you sometimes don't know if they were used
incorrectly and should actually have been a binary prefix.

By implementing a policy of using the decimal and binary prefixes
correctly, we can ensure that such ambiguity is removed in FDP output.

Not using the correct prefixes is like not bothering to spell correctly.
You might be perfectly understood almost all of the time, but that isn't
really the point.

-- 
dwmw2