home *** CD-ROM | disk | FTP | other *** search
- Path: nmo.gtegsc.com!gtenmo1!dkarr
- From: dkarr@nmo.gtegsc.com (David Karr)
- Newsgroups: comp.lang.c++
- Subject: Re: SPARCompiler problem
- Date: 09 Feb 1996 12:11:27 -0800
- Organization: GTE NMO
- Sender: dkarr@cheetos.nmo.gtegsc.com
- Message-ID: <uyd97ouumo.fsf@cheetos.nmo.gtegsc.com>
- References: <311A8705.41C6@llnl.gov>
- Reply-To: dkarr@nmo.gtegsc.com
- NNTP-Posting-Host: cheetos.nmo.gtegsc.com
- In-reply-to: Mark Spruiell's message of Thu, 08 Feb 1996 15:28:05 -0800
-
- >>>>> "Mark" == Mark Spruiell <mes@llnl.gov> writes:
- In article <311A8705.41C6@llnl.gov> Mark Spruiell <mes@llnl.gov> writes:
-
- Mark> I've encountered a discrepancy between two C++ compilers. I'm using
- Mark> both SPARCompiler C++ 4.0.1 on Solaris 2.4 and SGI C++ 3.2.1 on IRIX 5.2.
- Mark> The code below illustrates what I'm trying to do. Using an overloaded
- Mark> operator new(), I get unexpected results when using SPARCompiler, and
- Mark> the expected results with SGI C++.
-
- Mark> When this program is run on the Sun, the resulting output is
-
- Mark> ** C++ new
- Mark> A constructor
- Mark> A constructor
- Mark> A constructor
- Mark> ** My new
- Mark> A constructor
-
- Mark> I was expecting to see three constructor messages after "My new", which
- Mark> is what I get on the SGI.
-
- Just for your information, using CenterLine C++, I get the behavior
- you get from the SGI. It does seem like the SparcWorks behavior is
- incorrect.
-