Re: [boost] concept checking performance hit

From: Jeremy Siek (jsiek_at_[hidden])
Date: 2001-09-19 15:41:32

Weird, function_requires() should be dead code. I'm surprised g++ didn't
get rid of it. You might experiment with adding the "inline" keyword to
function_requires() in concept_check.hpp.

On Wed, 19 Sep 2001, Eric Ford wrote:
eford> > Is there a way to have a custom concept checking class check that a
eford> > class have a member function that has been templatized to take an
eford> > arbitrary type?
eford>
eford> I noticed that some of my code takes a noticable performance hit when
eford> I add function_requires<>().  At first I thought maybe I wasn't
eford> creating a concept checking class optimally, but AssignableConcept<>
eford> also causes a similar performance hit.  I'm using g++ 2.96 and have
eford> tried -O*.  Using -S to look at the assembly, the function calls to
eford> function_requires stay in the code.  Does anyone know if there's a
eford> work around for this?
eford>
eford> Thanks,
eford> E
eford>
eford>
eford> Info: http://www.boost.org  Unsubscribe: <mailto:boost-unsubscribe_at_[hidden]>
eford>
eford> Your use of Yahoo! Groups is subject to http://docs.yahoo.com/info/terms/
eford>
eford>

----------------------------------------------------------------------
 Jeremy Siek                          http://php.indiana.edu/~jsiek/
 Ph.D. Student, Indiana Univ. B'ton   email: jsiek_at_[hidden]
 C++ Booster (http://www.boost.org)   office phone: [phone]
----------------------------------------------------------------------