Re: [boost] Geometry/Vector Lib

From: Geoffrey Irving (irving_at_[hidden])
Date: 2006-09-22 16:10:42

On Fri, Sep 22, 2006 at 10:06:35PM +0200, Marcus Lindblom wrote:
> Sohail Somani wrote:
> >> -----Original Message-----
> >> From: boost-bounces_at_[hidden] 
> >> [mailto:boost-bounces_at_[hidden]] On Behalf Of Martin Bonner
> >> Sent: Friday, September 22, 2006 6:52 AM
> >> To: boost_at_[hidden]
> >> Subject: Re: [boost] Geometry/Vector Lib
> >>
> >> ----Original Message----
> >> From: Sohail Somani 
> >>
> >>> Marcus Lindblom wrote:
> >>>> I've sometimes done just:
> >>>>
> >>>>    class vector3d
> >>>>    {
> >>>>    public:
> >>>>       float x, y, z;
> >>>>
> >>>>       float operator[]( int index ) const
> >>>>       { return (&x)[index]; }
> >>>>
> >>>>       /* ... */
> >>>>    };
> >>> Won't that not work in the presence of padding? 
> >> In theory, yes.  In practice, I can't see why any compiler 
> >> would bother
> >> adding such padding.  
> > 
> > Theory is what holds the C++ world together! I wouldn't want to
> > knowingly invoke UB unless I can prove its worth my while (for example,
> > in the case of hardware specific optimizations).
> 
> You can add pretty easily some tests to see if it holds true for the 
> compiler you're running on.

Or you can just encode the test into an assertion into the class, as in
the code I posted.  Especially if you can make it a static assert (which
I'm not sure of).  It isn't theory if the code only compiles if it works.

> But small vector libs is such a matter of taste and convenience, just 
> like coding style. No solution will please everyone.

Geoffrey