Timeline for Is premature optimization always bad?
Current License: CC BY-SA 2.5
6 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| May 11, 2011 at 14:41 | comment | added | Cercerilla | @Malfist, but where do you draw the line? The whole "profile before optimizing" mantra seems tautological when you simply stop considering anything you'd do before profiling to be optimization. | |
| May 11, 2011 at 13:20 | comment | added | Malfist | @CodeninjaTim, I don't think he said, "pay no mind to performance while writing code." He said to profile before optimizing. There is a difference. For example, I know not to write a loop that does everything twice, but is that optimizing if I write it so it doesn't? I don't think so. | |
| Feb 8, 2011 at 19:05 | comment | added | Cercerilla | This is absolutely wrong. Profiling is a great tool, but it's entirely reasonable to say that you know what the critical path of your code is going to be, and to optimize that to an extent as you are developing. Granted, at first pass this should be mostly relegated to design and algorithmic concerns, and the low-hanging fruit of implementation level optimizations, but it's quite naive to just go about writing without a care in the world for performance until after you start profiling. | |
| Feb 8, 2011 at 17:24 | comment | added | Maglob | @Scott Whitlock: Good for you :) I did mention "or analyze". At least you took some measurements before and after to verify results, which is good. And if you knew (?) from previous experience (on that software/screen) that 60 seconds is way too much, you did knew that a) you do have a problem and b) where it roughly is. | |
| Feb 8, 2011 at 17:09 | comment | added | Scott Whitlock | Well, I once had this screen someone else wrote that tool about 60 seconds to open. I optimized it so it took less than 2 seconds. I didn't need to profile anything. | |
| Feb 8, 2011 at 17:04 | history | answered | Maglob | CC BY-SA 2.5 |