All of them. Because inevitably you're going to have a user on IE6 on Windows XP that the app doesn't work right for and you'll end up supporting it because management doesn't want to lose a customer. Ideally you'll want to be working with IE9+, Chrome ~48+, Firefox ~40+, Safari 9+ to ensure modern compatibility.
This is where clearly specifying a browser support list at the beginning of a web project (or regularly updating one for long projects) helps to set those expectations.
We have a list updated quarterly based on browsers with 5%+ market share that is baked into any requirements that clients sign off on, with additional time/money required for supporting anything outside of the list.
Saved our bacon many times when it turns out the client is using a 9 year old Firefox browser they refuse to upgrade, or it turns out one particular branch office is still stuck on IE8 (true stories...)
Completely agree with this, we tend to state that we will support the latest version and two major previous versions and if any other support is needed we can accommodate that at an extra cost which some of our clients have refused to pay.
For development purposes I use Chrome for work stuff and Opera for personal projects.
I am a professional webdeveloper for many years and IMHO the best way is to look in your tacking tool and test in everything which has more than say 5% of your traffic in the last months. Browser usage varies strongly from the type of customer you have.
Firefox, because I like it as a browser. Plus its Javascript error messages are often more illuminating than Chrome's.
Also, cross-browser compatibility is important to me, and I figure all my colleagues test stuff in Chrome and will complain about stuff that doesn't work there. I can't tell you how many Firefox-specific bugs I've found in our site.
The tough part is finding someone who will regularly check stuff out in IE or Edge, if you don't want to be spinning VMs up and down all day every day.
Chrome as it has excellent dev tools and a large market share. Then manually check in Safari, Firefox, iOS Simulator during development. Then BrowserStack for quick spot checks. Then check on real iOS/Android/Windows devices in our device lab as emulators are bad at replicating the true user experience (animations, click areas, godawful plastic trackpads on cheap Windows laptops, etc).
Cross-browser testing is hell and responsive has made everything multiple times worse for the amount of manual grunt work that has to go into every non-trivial custom web project. Greatly envy non-web developers who can build an interface once and it's done.
Firefox Developer edition - it's dev tools are amazing and rival or surpass Chrome for anything with the possible exception of the performance tab, which Chrome still has a edge in.
Then, using your analytics package (which you have, correct?) You run testing either on actual hardware, or something like Browserstack for everything else in descending order of popularity by unique visitor.
Really, you have several different rendering engines and JS engines in play at any given time - and it's helpful to have an agreed upon support model in place prior to starting development. Generally speaking, support anything with 5+% in the past 90, and make a argument for not spending time supporting things below that without very good cause.
I’ve always used Firefox and enjoyed Chrome for a bit due to coworkers claiming how much better it was but after Firefox Developer Edition I went back and still haven’t looked back. Unless I need to test Chrome specifics. Definitely Firefox Development Edition with Firefox Containers is the best combination.
13 comments
[ 3.4 ms ] story [ 113 ms ] threadWe have a list updated quarterly based on browsers with 5%+ market share that is baked into any requirements that clients sign off on, with additional time/money required for supporting anything outside of the list.
Saved our bacon many times when it turns out the client is using a 9 year old Firefox browser they refuse to upgrade, or it turns out one particular branch office is still stuck on IE8 (true stories...)
For development purposes I use Chrome for work stuff and Opera for personal projects.
Also, cross-browser compatibility is important to me, and I figure all my colleagues test stuff in Chrome and will complain about stuff that doesn't work there. I can't tell you how many Firefox-specific bugs I've found in our site.
The tough part is finding someone who will regularly check stuff out in IE or Edge, if you don't want to be spinning VMs up and down all day every day.
Cross-browser testing is hell and responsive has made everything multiple times worse for the amount of manual grunt work that has to go into every non-trivial custom web project. Greatly envy non-web developers who can build an interface once and it's done.
I did find this article[1] on using chrome and workspaces as an IDE of sorts. I think its kinda nifty.
https://y6326.wordpress.com/2016/02/17/basic-setup-chrome-de...
Then, using your analytics package (which you have, correct?) You run testing either on actual hardware, or something like Browserstack for everything else in descending order of popularity by unique visitor.
Really, you have several different rendering engines and JS engines in play at any given time - and it's helpful to have an agreed upon support model in place prior to starting development. Generally speaking, support anything with 5+% in the past 90, and make a argument for not spending time supporting things below that without very good cause.