zebra barcode printer in vb.net 7: Code Optimization with Mike Lee, the World s Toughest Programmer in Objective-C

Printer QR Code JIS X 0510 in Objective-C 7: Code Optimization with Mike Lee, the World s Toughest Programmer

Worse, it s likely such optimizations make no difference to your application s performance. An application is only as fast as its slowest part, and even trivial applications contain a mindboggling web of parts. Shaving off picoseconds at random is like spitting in the ocean. You re much more likely to get in trouble than you are to hit a fish. Rather than worry about the efficiency of every line of code, worry about your own efficiency. Adding code complexity to an application without making it faster is a waste of time. It s much more efficient to write 100 percent of your code to be readable, logical, and terse, then go back and speed up the 3 percent that s actually slow. The code you do not write is a gift that keeps on giving. It doesn t cause bugs or get in the way of adding features. It doesn t take time to figure out later, and it requires no maintenance. Unwritten code is never broken by an operating system update, nor does it ever become obsolete. The best way to keep your code base fast, small, and future-proof is to use the vendorprovided frameworks. This code is written by platform experts who are better engineers than you or I. Let them deal with hardware-specific optimizations. Not only do you not need to worry about your code breaking, but you ll also get future performance upgrades for free. When Core Animation debuted at Apple s Worldwide Developers Conference in 2006, it obsoleted thousands of lines of painstaking animation code in the nascent Delicious Library 2. Although our code was arguably better than what Apple had demonstrated, our boss, Wil Shipley, knew better than to let hubris lead us down the road of rolling our own. We ripped out our code and spent the next year helping Apple nail down bugs in the new framework by filing reports with Apple s Bug Reporter (also known as RadarWeb). Our work paid off the next year, as shown in Figure 7-6. Delicious Library 2 won the Apple Design Award for Best Leopard Application. Apple demoed our app on the same stage where Steve Jobs had delivered the keynote speech days before. The application loaded, looking much like the shipping version, with its iconic woodgrain shelves. Then the sample library materialized in a gorgeous fade-in animation, and the entire room literally gasped. The funny thing is that we gasped, too. We had never seen anything like it before either, because we didn t do that. All we did was use the Core Animation framework. The sexy fadein that so impressed our peers was a gift from Apple.

While individual lines of codes almost always don t matter, the order those lines appear in almost always does. The efficiency of the algorithm the code implements has a significant effect on its performance. Before I broke the bad habit of code-level optimization, I wrote a na ve array comparison based on the assumption that sorting the arrays before comparing them would take longer. I could have tested that assumption by writing it both ways and timing them, but it s much more efficient to familiarize yourself with the efficiency of basic algorithms. Sorted arrays can be compared in O(n log n), while my unsorted solution was O(n2). This means that if the arrays each contained 10,000 items, the na ve algorithm would require 100,000,000 operations, while the sorted algorithm would require 133,000 operations. With one million items, the better algorithm will perform eighty million operations, while the na ve algorithm will cause your computer to burst into flames on its way to one quadrillion operations.