Re: [boost] Re: Re: Re: Re: New patch for operators.hpp uploaded

From: Daniel Frey (daniel.frey_at_[hidden])
Date: 2002-10-24 04:08:08

David Abrahams wrote:
> 
> Daniel Frey <d.frey_at_[hidden]> writes:
> 
> > This comment applies to the gcc-3.0.4, too. At least I'm not
> > biased... ;)
> 
> Yeah y'are. Look at the progression for gcc again. Things got *worse*
> with the 3.0 series before they fixed it for 3.1 or 3.2.

Seen the smily? I know I'm biased as the GCC is the main compiler I'm
using and thus the only compiler I personally care about. Still I know
it's important to know about other compilers and their strengths and
problems. I think everyone who uses the GCC still prefers the 2.95.3 for
production stable code (at least you know its bugs :) and the GCC 3.2 is
slowly approaching a state where it could be taken into account. Of
course this is just my personal opinion, YMMV. When writing code for
boost, there is a strong need for code that is not just portable wrt the
C++ standard, but also portable wrt to real compilers. Your work helped
me a lot for the NRVO and some other people posted even more results.
Thanks everyone!

Of course one shouldn't judge a compiler by a single feature or a single
bug, indeed I think the Intel compiler is a very good if you have a need
for highly efficient code. Boost has - by it's nature - a need for very
sophisticated compilers as boost uses almost every feature the language
provides. It's hard work to write code which works everywhere and I
think it's impossible to do so for a single person (though I'm pretty
impressed how many compilers you have installed :). For the patch of
operators.hpp, I tried to check the compilers I have, now I think it's
up to the *users* of other compilers to modify the config-files to set
BOOST_HAS_NRVO, right?

Regards, Daniel

--
Daniel Frey
aixigo AG - financial training, research and technology
Schloß-Rahe-Straße 15, 52072 Aachen, Germany
fon: [phone], fax: [phone]
eMail: daniel.frey_at_[hidden], web: http://www.aixigo.de