Re: [boost] mcs::units informal review request

From: Noah Roberts (roberts.noah_at_[hidden])
Date: 2007-01-30 00:37:09

Matthias Schabel wrote:
>>> So far that seems to run true if you use the highest optimization  
>>> level
>>> for g++.  In VC++ I'm finding otherwise.
>>>
>> The cause is probably that quantity defines a destructor. It does  
>> exactly
>> what the compiler generated one does, so there is no reason to  
>> define it
>> explicitly.  Just comment it out.  msvc refuses to inline any  
>> function that
>> returns a type with a non-trivial destructor when exceptions are  
>> enabled.
> 
> Stephen,
> 
> Thanks for the suggestion - once I get my hands on a system with  
> Visual C++, I'll give your
> idea a try and see if this improves the performance. Interesting non- 
> feature, that...

I'll try it this friday; or maybe I'll stay late some day sooner but it
won't be tomarro.  I don't know that it is the issue though since the
original reason for trying to begin with was because mine was exhibiting
bad performance so I wanted to compare; as you know, I didn't get that
far because of the bug I posted about earlier (when I said in VC finding
otherwise I should have specified I was only able to test my own, which
shows same performance as yours with g++).

My quantity object has no user defined destructor so it wouldn't trigger
this kind of response.  I'm almost positive there are some optimization
switches that need to be turned on and both of us will want to document
this for our libraries.  I'm hoping that the compiler isn't too confused
by the metaprogramming to be able to run its optimizer over it.