Web Browsers for Testing

There have been three short periods since the creation of the web when you could get away with only using one browser to check your web pages. When the web was first created those initial web paes were only able to be viewed in the WWW (now renamed Nexus) browser. For a short period at the end of the 1990s Netscape 3 had so much of the browser market that you could ignore everything else. Then for a couple of years around 2005/6 Internet Explorer 6 similarly had so much of the market you could ignore everything else.

The rest of the time there have always been two or more different browsers with a significant market share and the time when the different browsers will all process a web page exactly the same way is still somewhere in the future. This means that there is no way to avoid the need to do at least some testing in different browsers so as to ensure that your pages are at least useable for the majority of your potential visitors.

With most browsers installing multiple versions of the same browser onto your computer is relatively easy. Only Internet Explorer complicates things for you with regard to running multiple versions and even that situation is much improved with virtual machine images available that can be run on Windows XP and the ability to run multiple versions of IE in XP mode on Windows 7.

With the two computers that I use I have 17 different browsers on one computer and 9 on the other with only two browsers in common between the two. That gives me 26 different browsers I could test with if I need to. Mostly these browsers are there because I prefer to install new browser versions separately and gradually transfer across when there are full version upgrades of a browser rather than upgrading in place. That way I have the prior version to use if the new one gives me problems plus for a while there will be a lot of people using both versions and so testing in both versions for a short while after the new one comes out is essential. With most browsers almost everyone upgrades soon after a new version comes out so once a new version has been out for a couple of monts you can forget about the old version. I just can't see any reason for deleting those old browsers.

The only browser where everyone didn't upgrade quickly to the next version is Internet Explorer. At the time of writing this there are three versions of Internet Explorer with a significant number of users and each works differently enough from the others to treat them as three different browsers. With all the other browsers the changes between versions are usually not all that great and almost everyone is running the latest version so we really only need consider the latest version of those browsers.

Now depending on what sort of site you are building and what you are using to build it with you may not need to test on every possible browser. In fact with thousands of different browsers out there no one can possibly test a page on all of them. What we can do is to prioritise the browsers based on how many people use them and how similar they are likely to be to other browsers. That gives us a number of different "layers" for our testing and so provided you start with the browsers at the top of the following list you need only go as far down the list as you decide you need to in order to do the amount of testing you feel is necessary based on your requirements. If you are using an open source script to generate your site and all you are doing is entering plain text then testing the firat couple of layers to confirm that the script is installed and working correctly and then picking one browser to enter your text is about all you need. If you are writing your own CSS and JavaScript then testing at least the first three or four layers may be needed.

The most important browser to test your pages with is the most popular browser. The browser which currently has more than double the user base of any other web browser (considering the versions of IE as separate browsers) is Firefox. That makes Firefox the most obvious choice as the place to start your testing. Testing with Firefox first not only means that you know the page works for the biggest percentage of users from the first test but also since Firefox is a relatively standards compliant browser most of what works in Firefox will also work in other standards compliant browsers and you'll likely only need to make changes for old versions of IE if any changes are in fact needed.

The next group of browsers to consider is a group of four browsers which when you add their userbase to that of Firefox means that you have over 90% of all your visitors covered. Those four browsers at the time of writing are IE8, IE7, IE6, and Google Chrome. This group not only provides you with the greatest percentage coverage of all your visitors, it also covers those browsers which are least compliant with the standards. If your page works with all five of these browsers then the chances are fairly good that it will work with most of the other browsers since most of the rest either comply with the standards or try to emulate one of the above browsers.

There are four different rendering engines used by those browsers with any significant share of the browser market and the browsers we have covered so far cover three of those rendering engines (ghecko, trident, and webkit). The fourth significant rendering engine is Presto and the main browser using that rendering engine is Opera. Opera is possibly the most standard compliant of the popular browsers and also provides te greatest functionality for being able to make it even more standard compliant and so is the perfect choice for ensuring that the code will work in standard compliant browsers. Other browsers with a similar level of popularity are Mozilla and Safari so you may also want to consider testing those browsers as well but since they are fairly similar to Firefox and Google Chrome respectively testing in those is not so important.

If you have tested in all the browsers listed so far then you have over 98% of your visoitors covered. Most likely if you look at your browser stats you will be unable to find any other browsers with a significant enough market share to make that browser worth testing. Any further browsers that we select for testing with will therefore not be chosen because of their popularity but rather will be selected in order to test for specific things. Many of you will decide that you do not need to test to this level since it really depends on exactly what it is you decide to test as to what browsers you need to install and test with at this point.

One browser which I do have installed and occasionally test with is a text only browser called Lynx. This particular browser can only display text and so in place of each image in the web page the alt text is displayed instead. This browser gives me a good idea of how the page will appear to those using either text only browsers or who are using a text reader where the text I see on the screen in Lynx is what will be read out to those listening to their web reader. It isn't perfect for checking web readers since they may also read out other information based on the tags in the page but without going to the trouble of actually installing web readers it at least gives me an indication.

The other browser which I have checked web pages in a couple of times is a webTV emulation which allowed me to see how much of my web pages are visible when viewed on a TV screen.

A professional web designer/developer needs to run all these tests to make sure that their web pages that they create will work properly in all the places where they are likely to be used. Any exclusions from this need to be explicitly specified in the contract before work starts so that the client knows where their pages are not guaranteed to work correctly. For someone creating web pages as a social activity or hobby all of these tests will not be done because those people are more concerned with getting their content out there than they are in making sure tat their pages work. Also since they are likelty to be using code written and tested by someone else and only plugging in content the chances are that if one of their pages works across multiple browsers then the rest will too. I do recommend that those people do at least test their pages in the most popular browsers first just to check that their setup is working correctly even if their subsequent testing is limited to the one browser that they themselves have chosen to use for their web accesses.


This article written by Stephen Chapman, Felgall Pty Ltd.

go to top

FaceBook Follow
Twitter Follow