|
Boost : |
From: Howard Hinnant (hinnant_at_[hidden])
Date: 2003-12-03 12:15:34
On Dec 3, 2003, at 10:29 AM, AlisdairM wrote:
> Well, I'm trying to use dimension in a non-metaprogramming use-case <g>
>
> We have quite a few places in our codebase where we allow the compiler
> to
> determine array size for us:
>
> sometype var[] = { initialiser-list };
>
> Often we need to pass this array to some function or library
> algorithm, and
> then we need the upper bound. This is not readily available from var
> without resorting to
>
> sizeof(var) / sizeof(var[0]);
>
> Much nicer to pick it out with
>
> dimension( var );
>
> So it's not something that can't be worked around, but going directly
> to
> the answer rather than seeing the calculation and worrying about
> whether
> packing / alignment is an issue etc. simplifies code reviews etc.
>
> The problem here is obtaining the type of var. Using a function
> template
> allows for type-deduction, but does not yield a compile time constant.
I see. Thanks. The compile-time version of dimension would be more
clumsy for you:
dimension<decltype(var)>::value;
Too clumsy?
> Personally, I would rather use tr1::array for this <g> but again we
> have no
> syntax to deduce array size here so must fall back on traditional C
> arrays.
<shocking!> :-)
-Howard
Boost list run by bdawes at acm.org, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk