By John Gruber
OUTLIER: Hardcore quality clothing.
Safari 2.0 is only available if you’re using Mac OS X 10.4 or later; Safari 1.3 offers a very similar rendering engine, but is only available on Mac OS X 10.3.9. Like most browsers, Safari provides its version number in the user-agent string it provides when identifying itself to your web server — which means that by examining your server logs, you can tell which version of Mac OS X Safari is running on.
So, by looking at the daily logs for daringfireball.net over the last few weeks, we can see how many Safari-using readers were using Mac OS X 10.4, 10.3.9, and older versions of the OS.
The usual caveats apply:
These figures only regard Safari users. No other browsers were counted — to my knowledge, there’s no way to determine which version of the OS a browser like, say, Firefox, is running on.
Daring Fireball readers almost certainly do not comprise a representative sample of the overall Mac user base. These figures do not show how many Mac users have upgraded to Tiger; they only show how many Safari-using Daring Fireball readers have. (And they’re probably close to the numbers for Daring Fireball readers using other browsers, but that’s just my hunch.)
Web server log analysis is an inexact science.
That said, here we go. The numbers in the last four columns are the percentage of Safari users using that version of Mac OS X.
|Date||Total Safari Users||10.2.x||10.3.0- 10.3.8||10.3.9||10.4|
In November 2003, I did this same analysis a few weeks after Mac OS X 10.3 shipped. Here’s a graph of both adoption periods, starting with the day before each OS was released in stores:
The adoption rate for Tiger seems to be slightly less than that of Panther, but, considering the sloppy accuracy of server log analysis, I’m scoring it as a tie, for now.
The Omni Group’s Software Update Statistics site is chock full of interesting stats. For OS version, they’ve got 10.4 at 44.5%, 10.3 at 53.1%, and 10.2 at a measly 2.5%.
For CPU type, they have G3 at 5.9%, G4 at 70.9%, and G5 at 23.2%. Which, if you do the math, leaves Intel at zero, for now.