should abs() be 'uint'?

Article View: comp.std.c

Article #2141

#2141

From: Eyal Lebedinsky
Date: Sun, 27 Sep 1992 00:02

7 lines
852 bytes

I did not see discussion or find it in the usual books but I did have
programs fail due to the fact that (with 16 bit ints) abs(0x8000) is
0x8000 which is still negative.

If the function was 'unsigned int' then the result would ALWAYS fit into
the return type. Of course, 'a < abs(x)' may force the compare to be
done in a wider type on 16bit implementations.

Technical Details

Message-ID: <eyal.717552134@ise>

Path:

arc1-ca.pugleaf.net!sparky!uunet!newsgate.watson.ibm.com!news.ans.net!malgudi.oar.net!caen!batcomputer!munnari.oz.au!manuel!csc.canberra.edu.au!echo!eyal

Show Full Headers

Complete Article Headers

Newsgroups: comp.std.c
Path: arc1-ca.pugleaf.net!sparky!uunet!newsgate.watson.ibm.com!news.ans.net!malgudi.oar.net!caen!batcomputer!munnari.oz.au!manuel!csc.canberra.edu.au!echo!eyal
From: eyal@echo.canberra.edu.au (Eyal Lebedinsky)
Subject: should abs() be 'uint'?
Message-ID: <eyal.717552134@ise>
Summary: there is a problem with 0x8000...
Keywords: abs
Sender: news@csc.canberra.edu.au
Organization: Info Sci & Eng, University of Canberra, AUSTRALIA
Date: 27 Sep 92 00:02:14 GMT
Lines: 8