Subject: Re: sort(1) behavior?
To: NetBSD-current Discussion List <>
From: Steven M. Bellovin <>
List: current-users
Date: 04/11/2004 16:41:00
In message <>, "Greg A. Woods" writes:

>Hmm...  It does seem that I was confused over the use of "radix
>character" in the above quote from the standard.  I've since found the
>POSIX definition of this phrase:
>   Radix Character
>   The character that separates the integer part of a number from the
>   fractional part.
>I was confusing this strange POSIX definition of "radix character" with
>something more in line with the traditional meaning of "radix" combined
>with the word "character" in this way, i.e. "an identifier specifying
>the base of the numbering system", e.g. "D" for "decimal", "H" for
>hexidecimal, "O" for octal, "B" for binary, etc.  I should know by now
>never to assume anything about what a standards document means when it
>uses what I would consider a common phrase.  :-)
>(I'm guessing the POSIX folks decided they had to choose some other
>phrase than "decimal point" since that one might be too confusing for
>those locales which don't use a "period or full stop" as the character
>indicating the "decimal point" even though the phrase "decimal point"
>should be totally unambiguous regardless of what character any given
>locale might use to indicate the decimal point.  Even Webster's 1913
>edition clearly defined the phrase "decimal point", as do all of my more
>modern dictionaries, though of course since they're all English language
>dictionaries they describe this separator character only as "a dot or
>full stop", but I'm sure any more international dictionary would include
>other characters used by other locales.)

The more usual phrase is "radix point".  Of course, unless they're 
planning on permitting hex, octal, or binary fractions, there's no 
reason not to use "decimal point" -- but maybe they do permit those...
>This is starting to split hairs since the example originally posted
>didn't use any non-zero fractional parts, but I still don't see anything
>there which mentions any significance for the decimal fractions, but
>that might just be my bias for integers showing through.  :-)


		--Steve Bellovin,