draft proposal for ref counting in D

[Walter Bright]

Michel Fortin wrote:

Le 25-juin-2013 à 21:40, Steven Schveighoffer a écrit :

 > On Jun 25, 2013, at 9:31 PM, Michel Fortin wrote:
 >
 >> Not at all. Autorelease pools were useful at a time before ARC so you 
wouldn't have to think of releasing manually every object called functions were 
returned to you. Instead, most functions would return autoreleased object and 
you'd only have to retain those objects you were storing elsewhere.
 >
 > Having used MRC, I appreciate what autoreleasepool did, but I thought of it 
being also as a kind of blanket way to allow the compiler to remove extra 
retains/releases in ARC.
 >
 > Is it not advantageous to release a whole pool of objects vs. releasing them 
individually during execution?  All releases and retains are atomic, so I 
figured one could do some optimization when it's all lumped together.

I haven't done any benchmarking, but I'd have to assume it is more advantageous 
to just return objects retained since Apple went to great lengths to make sure 
this can happen even when the convention is to return autoreleased.

There's no question it also simplifies the compiler. It's much easier to reason 
about pairs of retain/release than retain/autorelease.

 > I find the autorelease pools very GC-like -- you don't have to worry who uses 
or forgets the reference, it's kept in memory until you don't need it.


The concept was truly great, no doubt about that.