How do you know which browsers to test your web application on? Ideally, you want to test in as many browsers (and operating systems) as you can. But realistically, when you’re a small test group (or a solo act, like me), with limited test hardware and only 24 hours in a day (only 10 of which I’m really willing to spend working, but I’m just lazy like that) — it can be tough to get a lot of browser coverage. In the test environment I deal with now, there isn’t even a Mac available, for example, in order to test on Safari – though I suppose now that Apple has released Safari for Windows, there’s a way around that problem (assuming their Windows version uses the same rendering engine etc. as the Mac version – anyone know?).
Sometimes your project will tell you “we only support these browsers” and then, obviously, those are the ones you worry about testing. Some specific web applications will be developed such that they only work with one or two particular browsers (a really bad idea, in my opinion, but there it is). That saves you a lot of time, though it sucks for users. But more often, you’ll ask the project manager about browser requirements and get some vague answer about “whatever is popular…” Ugh, nightmare! Do whatever you can to get the list narrowed down. Sometimes you can make a compromise to do in-depth testing on a few browsers, and quick and dirty checking on many others. Negotiate, negotiate, negotiate.
If you, like me, somehow are still expected to write the browser requirements yourself, first consider your audience/users. If you’re testing an intranet for a company that only supports IE6, for example, there’s your answer. Or maybe you’re testing a web site that caters to graphic artists and creative types – better think about Mac and Safari! Mobile browsing is, of course, a whole other animal that I can’t even begin to address here.
W3Schools has a list of browser statistics that can be helpful. Note that Firefox gains market share every month, while IE6 loses it (not necessarily to IE7 either). There’s part of your answer. The O’Reilly network recently published some updated statistics from their logs (here, and more later here). O’Reilly’s audience is probably not typical of the average Joe Blow, but the numbers are informative (and if your users are all software developers, you’re home free with those numbers).
On the IE6 vs. IE7 question, I still lean towards IE6 for testing. Yes, IE7 continue to gain market since it was pushed out with Windows Update. However, lots of people don’t like it (it’s a pretty radical change from IE6) and IE6 is still beating it in the numbers. Firefox 1.5 vs. 2.0 is more of a toss-up. I find them so close in look, feel, and functionality, that you may be safe to just pick whichever you prefer (I’d go with 2.0 myself). None of the statistics I can find even break out 1.5 vs. 2.0. This all may change when 3.0 comes out (possibly this fall?), if it’s radically different.
I don’t know enough about Safari for Windows yet to say whether you can sub that for Safari testing on a Mac. I think though, if you have a Mac available, you should go ahead and test on it (at least with Safari), even if it’s just a cursory examination. Other browsers (Opera, Flock, Netscape, ?) can be considered “niche” browsers. If you get a chance to test with them, great. But unless you know that a significant portion of your user base uses one of those niche browsers, you might better be served spending your time on the more widely used browsers.
On my current project, I’m testing in:
- Firefox (required, and I chose 2.0)
- IE6 (required)
- IE7 (optional, so I’m just doing quick checks as I have time)
- Safari (also optional, but I’ll probably try it with the Windows version since my project hasn’t bought a Mac yet)
Happy (?) browsing!