alt Hacker NewsI do not get this argument at all. A long time ago I ported a simple sudoku solver to Objective-C by using the foundation classes, like NSMutableArray. It was terribly slow. All those messaging sending just to do what should have been a single instruction (or less!) That’s when I realized that if you want speed in an Objective-C app, you really are going to reach for the C subset. The objective part is really good for building GUIs, but not for pure computation.
Replies
When I was writing apps at Apple for the low-integer-version-numbered iOSes, this is what we typically did (since we had a single-core CPU in the MHz and <1GB).
UI shell in UIKit Obj-C, over a C++ or CoreFoundation (C) business layer, talking directly to sqlite.
I haven't seen the source of Apple apps and frameworks in over 10 years now, but I hope for their sake a lot of it has moved to Swift by now.
If I were CFed I'd mandate 2026 as the Year of Claude Code Radar Burndown. Their backlogs are insane and Apple actually addresses maybe 5% of what it knows to be wrong in a given year. Make it 2% when a UI Refresh is mandated.
That is one advantage of C++ - you can use higher level features like generic data types/algorithms, member functions, constructors, destructors, and iterators and still have performant code. You can use object-orientation without having to heap allocate each object and pay the cost of "virtual" calls. (which many other object-oriented languages couple together)