If T[new] is the container for T[], then what is the container for T[U]?

[Robert Jacques]

On Sat, 25 Apr 2009 09:16:39 -0400, Andrei Alexandrescu  
<SeeWebsiteForEmail at erdani.org> wrote:

> Stewart Gordon wrote:
>> Andrei Alexandrescu wrote:
>>> It looks we can't make it with only T[]. We need a genuine container  
>>> type, and T[new] was suggested.
>> <snip>
>>  What do you mean by a "genuine container type"?
>
> One with capacity and probably value semantics.
>
> Andrei

First, capacity is only static (i.e. can be included in the array as you  
suggest) for free-list based allocation. Which is _only_ used by malloc  
and mark-sweep GCs. And I'm really hoping Leandro gives us something  
better than mark-sweep.

Second, if people want value semantics, they can already use .dup or a[] =  
... when needed. Adding value semantics to arrays seems to me like a  
recipe for unintended poor performance by new to D programmers (who don't  
know) and experienced ones (who make a hard to find typo) a like. (I can  
see that value semantics for static arrays might be useful i.e.  
foo(real[4] vec), but I don't see a major need for dynamic vectors to have  
value semantics and every reason they shouldn't, i.e. foo(real[]  
big_image). Counter examples are desired and welcome.)