Apple Completes Migration of Key Service to Swift, Gains 40% Performance Uplift(infoq.com)
infoq.com
Apple Completes Migration of Key Service to Swift, Gains 40% Performance Uplift
https://www.infoq.com/news/2025/06/apple-swift-migration/
14 comments
Big discussion on the original source (243 points, 12 days ago, 228 points) https://news.ycombinator.com/item?id=44172166
Why on earth did everyone start using “uplift”
Because when you work for a big company using trendy jargon is an acceptable substitute for actually getting things done.
English is pretty fluid. Like yourself I don’t love the buzzword-y feel of business slang (oh how we’ve destroyed the impact of “impact” with “impactful”; please slap me if I utter “ideating”), but without the flexibility to produce those horrors, we wouldn’t have delights like lit, hangry, and sus.
It's a great series of books by David Brin
The Uplift War
https://en.m.wikipedia.org/wiki/The_Uplift_War
The Uplift War
https://en.m.wikipedia.org/wiki/The_Uplift_War
probably to appeal to the pc bros.
I dunno a lot of migrations to X technology are pitched as valuable on one quality dimension. We improved performance by Y%! When software “quality” should really consider many criteria: correctness, performance, developer productivity, maintainability and others.
I’d rather hear about a set of tradeoffs in these sorts of articles (performance was critical so we traded Y for X) than just bragging about one dimension of improvement.
I’d rather hear about a set of tradeoffs in these sorts of articles (performance was critical so we traded Y for X) than just bragging about one dimension of improvement.
> I dunno a lot of migrations to X technology are pitched as valuable on one quality dimension. We improved performance by Y%!
I'd add that more often than not these performance impacts are not due to changes in programming languages/frameworks/libraries/whatever but because some architecture or algorithmic constraint is done under the scope of the migration.
I'd add that more often than not these performance impacts are not due to changes in programming languages/frameworks/libraries/whatever but because some architecture or algorithmic constraint is done under the scope of the migration.
It’s a migration from Java to Swift. For some reason that key information was missing from the title.
TLDR: Native language wins over JVM optimisations.
In my recent but limited experience of working on iOS apps, they do need any performance boost they can get anywhere. Coming to the Apple ecosystem (appstore connect, testflight, xcode etc) it's quite a shock on how slow it all is (using xcode on a Mac Mini M4 for development, and as for web updates/approvals etc I now understand the pain others have mentioned here).
In my recent but limited experience of working on iOS apps, they do need any performance boost they can get anywhere. Coming to the Apple ecosystem (appstore connect, testflight, xcode etc) it's quite a shock on how slow it all is (using xcode on a Mac Mini M4 for development, and as for web updates/approvals etc I now understand the pain others have mentioned here).
I think you missed the 85% reduction in lines of code.
That sounds like a phenomenal reduction in complexity!
That sounds like a phenomenal reduction in complexity!
I wonder how long their build time is. The article from the “Things” folks said their system took 10 minutes to build despite being only 30k LoC, which seems really bad. I can’t help but think they must have done something off for it to take that long.
Hence, I wonder how long it take a to build what is probably a medium size system when done by a team at Apple.
Hence, I wonder how long it take a to build what is probably a medium size system when done by a team at Apple.