Date: 21 Feb 2002 12:06:08 -0800 From: swear@blarg.net (Gary W. Swearingen) To: Michael Wardle <michael.wardle@adacel.com> Cc: doc@FreeBSD.org Subject: Re: inconsistent use of data units Message-ID: <oaheoaa9cf.eoa@localhost.localdomain> In-Reply-To: <3C748CAA.9060805@adacel.com> References: <3C743707.3080505@adacel.com> <20020221003116.GA11893@hades.hell.gr> <3C744D39.1020308@adacel.com> <1014256250.304.66.camel@cocaine> <3C745639.8080509@adacel.com> <3C7463A5.5060204@pittgoth.com> <3C74673E.8010905@adacel.com> <4r664rbe2u.64r@localhost.localdomain> <3C748CAA.9060805@adacel.com>
next in thread | previous in thread | raw e-mail | index | archive | help
Michael Wardle <michael.wardle@adacel.com> writes: > No, the decision was made when the K, M, G, etc. prefixes were used. In > fact, I would suggest the only reason persons understood what GB or TB > meant (approximately) was the prefixes' SI roots (however loose). In > the same way, it is likely that larger sizes will be termed PB, EB, ZB, > YB, and so on, despite no computing body having suggested these or > approved them as meaning multiples of 1024. First let me say that you shouldn't assume that anyone here is unfamiliar with SI. Anyone getting a technical degree in at least the last thirty years should know the concepts (though IIRC, there was no SI thirty years ago). > It is not safe to assume that the prefixes (that bear an uncanny > resemblance to SI prefixes) mean multiples of 1024 (or more correctly, > multiples of 2^(10*n)) in the absence of any standard. It's not at all uncanny. It's a simple carry over of base-10 concepts (a sine qua non of SI) to the binary system. Prefixes are used to allow the easy naming of small-integer multiples of big "round" numbers. It's been choosen that it's 1000^n for decimal work and 1024^n for binary work. The choice of names for binary work was unfortunate, but the understanding of the prefixes is standard enough when discussing computer bytes and words, that it is safe enough to make assumptions about what the prefixes mean. People that use KB to mean 1000 bytes are just being confusing. > As I surely have said previously: > I have seen lots of different notations used for 1024 bytes, including: > - k > - K > - kBytes > - KBytes > - Ko > - ko > - KB > - kB > - kb > - KiB > - kiB > - KKB Right. And these never mean 1000 bytes, except in advertisements. I don't see a problem. Forbid most of them, and we WILL have the problem of having to fix them as they continually creep into the docs. > To call any of the above uniform practise would be untrue. Indeed, "KB" > is probably now the most widely used, but it is not by any means > uniform, unambiguous, or standardized, not to mention that the 1000 > v. 1024 issue is not addressed. But I do claim that it is uniform practise that all these terms refer to 1024 bytes (or other clear-from-the-context units like words), except "ko" and "Ko" which I've never seen. > It surprises me that when a well-respected engineering body *does* > propose a standard, that most persons I have talked to are in fact > against it! Well, we're still using designed-to-slow-typing QWERTY keyboards too. > Interesting. In common usage (as enforced by SI), kilo does indeed mean > 1000. A related analogy is that the IEEE proposal is merely ensuring > that "up" always means "up", instead of possibly meaning "sort of up but > a little bit sideways" in computing. Or that they are telling astronauts that they must use "up" to refer to "away from the earth" (because that's how the Geophysique Informatik defined it years ago) instead of the more natural "away from the floor". > I don't wish to blame, but perhaps it is the readers' mistake in > assuming that kilobyte meant 1024 bytes in the first place (again, this > is not a standard or even uniform in use). ... > Again, "kB" and "MB" would seldom be seen in this document, as I expect > in most cases 1024 is intended. The existence of references to KiB and > MiB would surely alert the reader that there was some terminology they > were not yet familiar with. When you expect that 1024 is intended, you make no mistake, despite your own statement to the contrary. Do you seriously think ANYONE doesn't expect that when discussing bytes? > Do not presume that every reader currently reads KB as 1024 bytes. That confuses two issues, but I will presume that KBytes as 1024 bytes, just like I'll presume that no reader will read KBytes as 1000 bytes. Will you use KBytes to mean 1000 bytes, or will you banish all use of SI prefixes on bytes? If you don't banish it, confusion will persist for a very long time, with readers always wondering if "KByte" means 1000 bytes or if the writer fell back on old habits. > Thanks for the interesting debate! :-) Yes. And I could easily adopt your side of the issues as you make reasonable arguments (too?). Things would have been much easier for many (including a certain spaceship to Mars) if we had all adopted SI units for everything (with an adjuct for binary system) long ago. The issues at hand are mostly practical ones, like those for the many gas stations in the US that paid for metric (or dual system) conversion in the 1970s only to have the trend fizzle out. Many of us see your "mibi" thing fizzling in the same manner. To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-doc" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?oaheoaa9cf.eoa>