[swift-evolution] What about garbage collection?

Austin Zheng austinzheng at gmail.com
Tue Feb 9 19:09:28 CST 2016


Your experiences may vary, but at the last large mobile company I worked for our Android engineers spent far more time wrestling with the GC than our iOS developers spent on similar memory issues. (A lot of this had to do with media handling or rendering complex UIs containing many high resolution image assets.)

That's not to say that Android apps can't be performant or smooth, nor that GC on Android hasn't greatly improved over time, but UI application development on a GC based resource constrained platform comes with its own tradeoffs. I would personally rather deal with breaking cycles (which can be worked around with the proper foresight and architecture) than placating a GC.

Sent from my iPhone

> On Feb 9, 2016, at 2:16 PM, Goffredo Marocchi via swift-evolution <swift-evolution at swift.org> wrote:
> 
> I generally agree that refcounting is not a solution taken by the language and tools developers at Apple for laziness, but if we are quoting the failures at using GC for UI heavy interactive apps... I do disagree with an example, but again that probably could only be sorted by a huge huge thread with better informed people: .NET vs WinRT was not a historically purely just performance and fully merits driven battle, politics between the two different divisions that developed each alternative have to be considered (and lo and behold the solution pushed by the Windows team won ;)).
> 
> Also... we are ignoring a large green electronic elephant in the room using Java and not doing. A bad job with very similarly priced handsets... :). Sorry for the noise though and thanks for all these informative posts :).
> 
> Sent from my iPhone
> 
>> On 9 Feb 2016, at 17:23, Joe Groff via swift-evolution <swift-evolution at swift.org> wrote:
>> 
>> Microsoft tried and failed several times to reinvent their stack on top of .NET, and has since retreated to a refcounting-based foundation for WinRT
> _______________________________________________
> swift-evolution mailing list
> swift-evolution at swift.org
> https://lists.swift.org/mailman/listinfo/swift-evolution


More information about the swift-evolution mailing list