|
@Catfish_Man | |||||
|
.@tapbot_paul oh, plist encoding/decoding should be a bunch faster due to various changes, but I’ve had trouble getting reliable numbers.
|
||||||
|
||||||
|
|
David Smith
@Catfish_Man
|
13 Jun 16 |
|
btw, NSArray and CFArrayCreate* will now return _NSArrayI and _NSArrayM instances on iOS/watchOS/tvOS as they have for some time on macOS.
|
||
|
||
|
Paul Haddad
@tapbot_paul
|
13 Jun 16 |
|
favorite iOS/macOS 10/.12 performance tweak?
|
||
|
|
||
|
|
David Smith
@Catfish_Man
|
13 Jun 16 |
|
os_unfair_lock by a light year. In F/CF land though, there’s a lot… hm…
* CF_IS_OBJC is way faster (used in bridging)
|
||
|
|
||
|
|
David Smith
@Catfish_Man
|
13 Jun 16 |
|
* _NSDictionaryM combines buffers for 33-50% fewer mallocs
* -retain/isEqual/hash on CF types calls new _CFNonObjC*
|
||
|
|
||
|
|
David Smith
@Catfish_Man
|
13 Jun 16 |
|
* CFRetain calls objc_retain instead of -retain
* Collections call objc_retain instead of CFRetain
* Silly memcpy tricks
|
||
|
|
||
|
|
David Smith
@Catfish_Man
|
13 Jun 16 |
|
remind me to dig up some more later. Out and about now.
|
||
|
|
||
|
|
David Smith
@Catfish_Man
|
13 Jun 16 |
|
and CFStringCreateCopy early-outs for tagged pointers now, which makes it 4x faster for ~40% of strings.
|
||
|
|
||