You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I'm quite amazed by how fast DifferenceKit is compared to Apple's Foundation diffing.
Can someone please explain what's the catch here?
Is there any reason why Apple chooses to use slower algorithm for their collection diffing?
Here's a sample test I run to compare results:
for _ in 0..<5{varsource:[Item]=[]
for i in 0..<10_000{
source.append(Item(value: i))}lettarget:[Item]= source.shuffled()measure(name:"DifferenceKit"){let _ =StagedChangeset(source: source, target: target)}measure(name:"Foundation"){let _ = target.difference(from: source)}}
Output:
Time: DifferenceKit - 29.129458333500224 msec
Time: Foundation - 1961.1885833332963 msec
Time: DifferenceKit - 17.962249999982305 msec
Time: Foundation - 1882.0110833335093 msec
Time: DifferenceKit - 17.888999999740918 msec
Time: Foundation - 1904.927041666724 msec
Time: DifferenceKit - 17.906958333242073 msec
Time: Foundation - 1907.209416666774 msec
Time: DifferenceKit - 18.071041666644305 msec
Time: Foundation - 1919.3862499996612 msec
The text was updated successfully, but these errors were encountered:
Thanks. I'd say the performance tuning of CollectionDifference is quite amazing as same as DifferenceKit, but fundamentally they are based on different algorythms. CollectionDifference is based on Myers's algorythm which is used for general purposes such as text diffing, and DifferenceKit is based on Paul Heckel's which is more suited for UI.
In evidence, CollectionDifference is remarkably fast compared to other libraries which is based on Mayer, but it's not even close to DifferenceKit.
For more information: https://github.com/ra1028/DifferenceKit#comparison-with-other-frameworks
Checklist
I'm quite amazed by how fast
DifferenceKit
is compared to Apple's Foundation diffing.Can someone please explain what's the catch here?
Is there any reason why Apple chooses to use slower algorithm for their collection diffing?
Here's a sample test I run to compare results:
Output:
The text was updated successfully, but these errors were encountered: