iOS JSON Parser Comparison

May 15, 2012 Brian Croom

This article was first composed by Peter Raboud

Multiple methods of parsing JSON data are available to iOS developers, but it is not clear which of these libraries is the best option. We investigated this problem by benchmarking the performance of several JSON libraries for iOS:

The tests were done by parsing several different-sized sample data multiple times with each of the libraries, and measuring wall time and CPU time as performance metrics. The tests were run on a 4th generation iPod Touch running iOS 5.1 (9B176). Since wall time and CPU time are strongly correlated, with a correlation coefficient of 0.999999, we did not present the CPU time data.

From testing, we discovered that the native method performed the best. The next fastest library is YAJL, which took about twice as long as the native library. Both of these methods only required a single line of code to use, and the native method is even easier, since it doesn’t have any dependencies to import.

One potential issue that we discovered is that the YAJL library can cause issues if using ARC. Due to changes in ARC, weak and assign are now mutually exclusive properties, which breaks some of the code in YAJLParser.h. Line 152 reads “@property (assign, nonatomic) _weak id delegate;“. Simply deleting weak seems to fix this problem. While this issue has been open on Github for 7 months at the time of writing, this workaround is the only solution yet discovered, and the author has yet to respond to this issue.

To conclude, the native library is the best choice as far as performance goes, but it can only be used in apps that exclusively target iOS 5.0+. For apps that must provide legacy support, YAJL is the best choice.

About the Author


Mastering Mobile Web
Mastering Mobile Web

Creating a mobile website can be challenging. Here’s a simple breakdown of how to make your mobile website ...

Android Roundup: Week of May 7, 2012
Android Roundup: Week of May 7, 2012

Cool things this week: Analytic options: use Google Analytics, Flurry, or Omniture. There’s also a new too...


Subscribe to our Newsletter

Thank you!
Error - something went wrong!