Why does D rely on a GC?

[Timon Gehr via Digitalmars-d]

On 08/19/2014 07:11 PM, bachmeier wrote:
> On Tuesday, 19 August 2014 at 16:17:02 UTC, Dicebot wrote:
>> On Tuesday, 19 August 2014 at 15:16:31 UTC, Ary Borenszweig wrote:
>>> Also, the list seems way too big. It's ok from a purist point of
>>> view, to make the compiler nice and clean. But that's not a good way
>>> to make a fast compiler.
>>
>> This is why I won't bother investing any time in it for a few more
>> years at least. It may have really cool language features, big
>> development team and famous names behind it but in terms of compiler
>> maturity it is still a very long road to go until it gets even to DMD
>> capabilities.
>
> I won't look at it again for a different reason. They're the types that
> say "A monad is just a monoid in the category of endofunctors, what's
> the problem?"

Sure, so one should point out that problem may be made out to be that 
the monoidal product [1][2] is underspecified for someone unfamiliar 
with the convention (in this case it should be given by composition of 
endofunctors, and the associator is given pointwise by identity 
morphisms). (But of course, the more fundamental problem is actually 
that this characterization is not abstract enough and hence harder to 
decipher than necessary. A monad can be defined in an arbitrary 
bicategory... :o) )

What do /you/ think is the problem?

>  but they're serious.

I have a hard time believing this. While it is true that those concepts 
are not complicated [3], it seems to be generally acknowledged that some 
carefully compressed characterization/definition alone is not usually 
the most helpful introduction of them, especially so if the application 
is going to be in a more concrete setting.

[1] http://en.wikipedia.org/wiki/Monoid_%28category_theory%29
[2] http://en.wikipedia.org/wiki/Monoidal_category
[3] But, quite apparently, they are sometimes hard to communicate.