Summation on HP 42S
|
09-28-2018, 12:33 AM
Post: #36
|
|||
|
|||
RE: Summation on HP 42S
(09-27-2018 10:37 PM)pier4r Wrote: That's impressive It's nice as long as it works... Free42 is at version 2.0.21a on iOS, and 2.0.21 on all other platforms, because the latest iOS SDK broke part of the number display logic, causing it to display unwanted leading zeroes in certain cases in HEX and OCT modes. This is code that has worked fine on more than half a dozen platforms, compiled with many different compilers, over the course of more than 13 years... and suddenly an optimizer bug in the latest iOS compiler caused it to break. Don't get me wrong, optimizers are great when they work, but when they don't, they can cause malfunctions in the weirdest places. Keep in mind that in the greater scheme of things, Free42 is a pretty simple piece of software, where bugs tend to be noticed quickly. I shudder to think what obscure optimizer bugs could do in really complex information systems! The person holding that talk apparently works for a high-speed trading company, where speed is everything. Good for him, but in the places where I have worked, and work today, I would rather disable all optimizations everywhere, and only enable them in those places where they make a significant and predictable difference. Having the compiler surprise me by going Mathematica on me and translating code to a closed-form equivalent strikes me as scary rather than exciting. |
|||
« Next Oldest | Next Newest »
|
User(s) browsing this thread: 4 Guest(s)