Unofficial wish list status.(Jul 2008)

[Robert Fraser]

Brad Roberts Wrote:
> Functional or not isn't a whole program decision (at least within D and 
> it's future as it seems to be unfolding).  Within most, if not all, 
> applications, there are reasonable subsets that are, or could be, 
> functional style.  Many are already, without language support to help 
> enforce the style or take advantage of the benefits.

The "benefits" of automatic parallelization can only be seen on a macro 
scale. For trivial functions, even if there's an array scan involved, it will most
likely be more overhead than good. Most imperative systems will need to
be redesigned if they want to take advantage of these things... not that
that's a bad thing. But once again, the question is: do the benefits
outweigh the costs?

> Everyone has their opinion, and aren't we glad for that privilege.  The 
> nice part is that const is optional.  There's only a very few places where 
> that leaks out, the obvious example being the phobos definition of 
> 'string' and that's avoidable.  The current implementation of phobos is 
> far from perfect (don't start that thread, please) in it's use of const, 
> but that's part of being an alpha version.  As it evolves, applications 
> and libraries should be able to choose to ignore const completely.  Any 
> part of phobos or the compiler that gets in the way of NOT using const, 
> should be examined quite carefully.

That's true... if you're one developer working on an application. But that's
not true if you're writing a library -- some of people won't use your library
if it doesn't support const. But, yes, in the end it's just a matter of opinion
on what the right language design is.