Re: Very concurrent garbage collector for ARC

So, if I manage to do that, how could I propose a patch to Apple?
Also, is there anyone on this list that could give me some guidance on how to build Apple's libobjc from source?

Em 30/09/2015, às 01:15, Jens Alfke escreveu:

>
>> On Sep 29, 2015, at 8:34 PM, Paulo H. Taka Torrens <paulotorrens@gnu.org> wrote:
>>
>> Well, this probably could be turned on/off using a environment variable (something like OBJC_CYCLE_GC)... but the point is that the overhead shouldn't be that big; Inferno's algorithm was projected to do really small steps each time, so that it wouldn't have to suddenly stop-the-world as most GCs.
>
> Which is cool; but it's still going to lower overall performance by some amount. (Back in the day, the rule of thumb was that any garbage collector would add at least ~10% overhead to the CPU time taken by a typical program.)
>
>> I'm still testing the runtime overhead on GNUStep, but I believe that the overhead will be really small. I would probably be able to test the same thing for Apple's as well (if I manage to build it from source…).
>
> That would be cool!
>
> —Jens
>
>

_______________________________________________
Do not post admin requests to the list. They will be ignored.
Objc-language mailing list (Objc-language@lists.apple.com)
Help/Unsubscribe/Update your Subscription:
https://lists.apple.com/mailman/options/objc-language/anirudha.k.cpp1%40blogger.com

This email sent to anirudha.k.cpp1@blogger.com