delorie.com/archives/browse.cgi | search |
From: | kevinx AT tst DOT hk DOT super DOT net (Mr Kevin Yeung) |
Newsgroups: | comp.lang.c,comp.os.msdos.djgpp |
Subject: | Re: DJGPP failed library tests? |
Date: | 2 Nov 1997 12:24:15 GMT |
Organization: | Hong Kong Supernet |
Lines: | 21 |
Distribution: | world |
Message-ID: | <63hrdf$mpt$1@tst.hk.super.net> |
References: | <639vjh$dru$1 AT tst DOT hk DOT super DOT net> <tgmEIyM5w DOT Mz AT netcom DOT com> |
NNTP-Posting-Host: | is1.hk.super.net |
To: | djgpp AT delorie DOT com |
DJ-Gateway: | from newsgroup comp.os.msdos.djgpp |
Hi, Try this: sqrt(-1.0); errno won't get set to EDOM. Also: assert(fabs(0.02 - 0.02) / 0.02 < 4 * LDBL_EPSILON); will fail. Did I miss something? Thanks a lot. Thomas G. McWilliams (tgm AT netcom DOT com) wrote: : Note that floating point tests often "fail" in the least significant : digits. You should examine the output of the tests and see in which way : that they failed. DJGPP's math library is highly regarded. Without : any specifics from you, it is impossible to answer you question. : Post your results. -- Kevin Yeung Change x to y in email address for private correspondence
webmaster | delorie software privacy |
Copyright © 2019 by DJ Delorie | Updated Jul 2019 |