Path: utzoo!utgpu!water!watmath!clyde!att!osu-cis!tut.cis.ohio-state.edu!bloom-beacon!WSL.DEC.COM!hania
From: hania@WSL.DEC.COM
Newsgroups: comp.windows.x
Subject: Re: Xtk "unsigned int" typedefs considered harmful
Message-ID: <8809131742.AA06973@gilroy.pa.dec.com>
Date: 13 Sep 88 17:42:56 GMT
References: <8809131658.AA03210@EXPO.LCS.MIT.EDU>
Sender: daemon@bloom-beacon.MIT.EDU
Organization: The Internet
Lines: 22


>> The point is that having Dimension, et.al., typedef'd as unsigned ints
>> has no redeeming virtue whatsoever

>Yes they do.  They are unsigned because things that are Dimensions are
>cardinal in nature.

And herein lies the problem: many people think of the C type "unsigned"
as "cardinal".  This is a mistake.  There is no "cardinal" type in C. 
The best way to think of "unsigned" is as a bit field; the only
examples of the use of "unsigned" that I can find in Kernighan and
Richie use them in that capacity.
In addition, there is no type-checking done on "unsigned" values, to
make sure that they are positive, as one would expect for "cardinal" types.

These are all arguments that "unsigned" should never be used in C to
mean "cardinal"; such use is, at best, misleading.  I believe that at
least some of the toolkit implementers believe that it was a mistake to
have things like Dimension typedef'd as unsigned.  The best the rest of
us can do is agree.

   Hania