Mailing-List: contact cygwin-apps-help AT sourceware DOT cygnus DOT com; run by ezmlm Sender: cygwin-apps-owner AT sourceware DOT cygnus DOT com List-Subscribe: List-Archive: List-Post: List-Help: , Delivered-To: mailing list cygwin-apps AT sources DOT redhat DOT com Date: Mon, 15 Oct 2001 11:21:21 +0200 From: Corinna Vinschen To: cygwin-apps AT cygwin DOT com Subject: Re: GCC-3 additions - related question Message-ID: <20011015112121.J1696@cygbert.vinschen.de> Mail-Followup-To: cygwin-apps AT cygwin DOT com References: <002901c154a1$03e8c9a0$0200a8c0 AT lifelesswks> <20011014203220 DOT 85319 DOT qmail AT web14510 DOT mail DOT yahoo DOT com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline User-Agent: Mutt/1.2.5i In-Reply-To: <20011014203220.85319.qmail@web14510.mail.yahoo.com>; from danny_r_smith_2001@yahoo.co.nz on Mon, Oct 15, 2001 at 06:32:20AM +1000 On Mon, Oct 15, 2001 at 06:32:20AM +1000, Danny Smith wrote: > --- Robert Collins wrote: > > > ----- Original Message ----- > > From: "Danny Smith" > > > > > > > > > Lately, I've been hearing this with STL code: > > > "This compile well using gcc! Why it not MS!" > > > Thats what I want -- the first part of that anyway. > > > > Danny, it sounds like you might know the answer to this: are there known > > issues for building native code using the cygwin gcc with C++ templates? > > > > Specifically, I've asked Chris on the developer list about using > > templates for some stuff (iterators particularly are getting real old > > for me), see > > http://www.cygwin.com/ml/cygwin-developers/2001-10/msg00018.html > > for the thread. > > > I've built STLPort, Bench++ (Stepanov benchmarking suite) and some of the > Boost package with mingw 2.95.3 and gcc's ability to build template code is > not a problem compared with what I've read for other windows compilers. > The issue is with efficiency. Efficiency of writing generic code may be > enhanced by templates. But... Debugging template code is a real pain -- > stange errors with instantiatiation of templates have to be tracked through > a maze of headers. And there is a runtime performance hit. I don't have > the Bench++ results handy but there was a significant penalty for using > generic template code compared to C code. I expect gcc-3.0 to be better, > at least as far as speed of compiled code. Speed of compilation, however, > is a problem when your testing. > > I would agree with Corinna. Another reservation was the observation in earlier gccs that using templates bloats the size of the executable/dll drastically. Is that still true? Corinna -- Corinna Vinschen Please, send mails regarding Cygwin to Cygwin Developer mailto:cygwin AT cygwin DOT com Red Hat, Inc.