Re: [boost] [xint] Boost.XInt formal review

Subject: Re: [boost] [xint] Boost.XInt formal review
From: Steven Watanabe (watanabesj_at_[hidden])
Date: 2011-03-09 14:42:47

AMDG

On 03/03/2011 09:33 AM, Stewart, Robert wrote:
> The use of imperative programming to control what should be distinguished at compile time is disturbing.  The runtime overhead may be provably insignificant, but I'd like to see that proved and documented.  For example, most functions contain "if (Nothrow) { ...} else { ... }".  That means there is a great deal of code carried in either case that never executes.  The branches also mean there is more opportunity for bugs to creep into one variant and not the other.  An application that only uses integer_t<options::nothrow>  instances, there is a lot of code in those else branches that isn't needed.  Far better would be for no_throw_integer to derive from integer_t.  Changing this aspect, if performance and binary size tests show it to be important, may affect the set of types or interfaces in the library, and must therefore be done before acceptance.

Frankly, I don't see why this is important.  I don't
really care how the library selects the correct
implementation.  The interface is fine from a
user's point of view.

In Christ,
Steven Watanabe