Re: Binary units

[Christian Rose <menthos gnome org>]

tor 2003-12-18 klockan 21.10 skrev Andreas J. Guelzow:
> > > 1 KB traditionally has been 1024 byte, we should not change that!
> 
> > My point being that just saying that "it's tradition" is as dumb as just
> > saying "it's wrong". 
> 
> Pardon me? There is a traditional/historic use of KB, MB etc.

Certainly. And if you would have read just a little bit further I think
I made it pretty clear that it's the making of such bold statements
without taking into account other issues that I'm criticizing.

But you unfortunately left out those parts.


> Moreover, none of these abbreviations are covered by SI since B is not
> an SI unit.

The discussion is about prefixes, not units. The SI prefix can be, and
are, used with other non-SI units too. kilocalories, kilobar, megatons,
kilowatt hours, kiloponds, and so on. There's an endless amount of
examples. Heck, they're even used in accounting. And various currencies
are definately not SI units.
So where did this novel idea that the prefixes can only be used together
with SI units come from?


> > Personally, I see the SI proposed prefixes slowly getting more
> > acceptance. It's still far beyond what one would expect, but it's slowly
> > crawling into usage in some places.
> 
> It may be crawling into use in some places, except counting storage
> space is not one of those areas.

Well, end of time will happen before storage hardware manufacturers will
use those prefixes, since it's obviously much more convenient that
consumers confuse their G (=10e9) with the common use of G as a binary
prefix in this area. But the SI proposed prefixes are certainly slowly
becoming more commonly used in software.


> > Also, the biggest complaint I always hear about free software is that it
> > just copies, never innovates. Free software people never has the courage
> > to do anything new, but rather just copies what is already there, and is
> > thus always playing catch-up with copying new ideas.
> 
> Well, just because something is new does not mean it can't be dumb.

True. It all comes down to the fact in how much one evaluates many
different aspects that may be important. Which brings us back to the
bold, unbacked statements like "it's wrong, end of story" or "it's
tradition, end of story" department.


> > Even though I believe that is to a large extent false, and that we have
> > probably been changing a lot of that which was true about this in GNOME
> > since the GNOME 2.0 days by having the courage to do some controversial
> > UI changes, I'm still puzzled anytime I see someone rejecting ideas or
> > concrete proposals just because it conflicts with a "traditional" way,
> > and just wants the discussion to stop there.
> > Come on, wake up, anything new is bound to conflict with a "traditional"
> > way, that's the very nature of doing anything new and being in the lead
> > of development.
> 
> We should hope that you have better reasons than just that it is
> non-traditional. 

The SI proposed prefixes were obviously not introduced for the fun of
it. The reasons for introducing them and their background history has
been documented and presented in a countless number of articles for many
years, ranging from scientific papers to ordinary computer magazines and
websites. Are you claiming you know nothing about the background?


> > And this is such an excellent example of an area where we could have had
> > the courage to share the lead of development, instead of cowardly just
> > refer to traditional use and leave it like that, just the behavior that
> > we wanted to change.
> 
> Perhaps if you would explain the benefit of using decimal powers to
> measure something inherently given in binary powers, you could persuade
> people.
>
> But so far the argument in favour of decimal powers eludes me.

Hmm, as I read the thread the argument was for using binary powers by
their nonambiguous name, not decimal powers. If someone suggested using
decimal powers I must have missed it.

Still, assuming you meant using the gibi names and friends instead of
the giga names and friends, the clue is in the claim of "inherently
given". In my experience, only computer enthusiasts and otherwise
technical people usually expect everything to be counted with binary
prefixes in this field. But those alone are not the target of GNOME, in
fact it is only a fraction of the total market for a desktop
environment.

And trust me, for an "ordinary user" it is often far from given that the
120 GB hard disk drive he or she ordered is in fact then only reported
as 111 GB total size by the computer. That's a whole 9 GB difference,
something that was more than the size of a fairly big drive not many
years ago.

That this "loss" exists is not only difficult to explain, it's also far
from expected, since people have had training all their life (well,
perhaps this is "only" true to areas outside the US and areas that have
only recently adopted the metric system) that prefixes are a decimal
thing. Entire societies have used that norm for centuries for everything
measurable in life. And suddenly we're placing an environment on these
users' desks which not only breaks all those conventions, but does it in
the most ambiguous way possible: by masquerading them as the ordinary
prefixes by using the very same names.


Christian