Re: [boost] Re: Re: Re: Re: Profiling Library suggestion

From: Beman Dawes (bdawes_at_[hidden])
Date: 2005-02-19 10:53:25

At 09:45 AM 2/19/2005, christopher diggins wrote:

 >Thanks for the advice Beman. I would like to submit the profiler library 
to
 >boost *without* a high-resolution timer, and rather only provide the
 >guarantee of satsifactory performance when used with boost::timer. I am
 >using a policy driven design and the user can supply an arbitrary timer
 >class which models boost::timer. I see timers as a separate problem from
 >the profiler, that I don't want to get into.

That seems like a reasonable idea to me. Eventually someone will get 
irritated enough at the lack of a high-resolution timer to work through the 
issues, but in the meantime you can be moving your profiling library 
forward.

 > What do you think, would this kill the chances of a profiler ever
 >becoming part of boost?

No way to tell for sure until there is a formal review, but unless it 
somehow compromises the design I don't see why anyone would object to that 
approach.

--Beman