Message-ID: <6QQ2pO8zccB@jocokko.edition.bonbit.org>
References: <5d3co5$o7p AT huron DOT eel DOT ufl DOT edu>
MIME-Version: 1.0
Content-Type: text/plain; charset=ISO-8859-1
Content-Transfer-Encoding: 8bit
From: kay AT edition DOT bonbit DOT org (Kay Hayen)
Subject: Re: GCC Bug - Information
Date: 07 Feb 1997 00:00:00 +0000
Newsgroups: comp.os.msdos.djgpp
Distribution: world
Lines: 45
To: djgpp AT delorie DOT com
DJ-Gateway: from newsgroup comp.os.msdos.djgpp


Hallo Daniel.

The bug you posted is only a violation of ANSI-specifications. The  
proprocessor shouldn't allow to compile it, but it does.

Some people call it a feature and you should see their point instead of  
getting angry. I my eyes is a compiler bug something that makes a legal  
source compile to an incorrect program.

The bug you posted is different. It allows you to do something, you were  
normally not allowed. This doesn't cause any harm, but enables people to  
code in a language that is not ANSI-C++, but is an extension of it. Some  
people call it a feature and I suppose, this is why it is a known bug,  
very likely *not* to be removed.

Concerning bugs I'd say that version 2.7.2.1 is far more bugfree than  
any other C++ compiler I used. I used gcc 2.6.3 for long time and found  
it very buggy, 2.7.0 was horrible!

I also used BC in some versions, but they were not too good. Remembering  
4.0 I'd say one can compare it to 2.7.0 of GNU, but no serious developer  
would switch to .0 versions anyway. I'll leave my hands off from gcc  
2.8.0 for long.

One thing that is very good about gcc is response time. From Borland one  
is very likely to have bugs never being corrected, whereas gcc has very  
good support, known bugs and other nice things.

One thing that is not so good is that features are being implemented  
only very slowly. I have a lib that compiles well with BC4.5 and many  
other compilers, but gcc saying "sorry not implemented".

        Mls, Kay










## CrossPoint v3.11 R ##