Re: [boost] [XInt] review

Subject: Re: [boost] [XInt] review
From: Chad Nelson (chad.thecomfychair_at_[hidden])
Date: 2011-03-09 10:52:35

On Wed, 9 Mar 2011 09:41:58 -0500
"Stewart, Robert" <Robert.Stewart_at_[hidden]> wrote:

>>> Why can the comparison operators throw? I would have expected that
>>> they could be implemented to never throw.
>>
>> So far as I know, the only reason they would ever throw is if
>> they're passed a Not-a-Number value, or if something
>> unpredictable has gone seriously wrong.
>
> I would hope you could be more definitive than that and clearly
> document the behavior.

I can be definitive about the Not-a-Number behavior, and anything else
that I put into the code. I can't be definitive about anything
unpredictable. Out-of-memory isn't a problem in the comparison
functions, but it's probably the most common example of stuff that can
go wrong unpredictably.

>> > magnitude_t:
>> > * It probably needs to be POD for you to make assumptions
>> > about the layout.
>>
>> The "struct hack" is a standard C idiom [...]
>
> A C struct is POD.

So far as I've been able to determine, a C++ struct with some
non-virtual functions, and that doesn't inherit from anything, acts
like POD in every case. There's no need for a vtable or any other
appended data that I've been able to imagine, and there certainly isn't
on the two compilers I've directly tested (MSVC on Windows and GCC on
Ubuntu Linux).
-- 
Chad Nelson
Oak Circle Software, Inc.
*
*
*