Related
As it currently stands, this question is not a good fit for our Q&A format. We expect answers to be supported by facts, references, or expertise, but this question will likely solicit debate, arguments, polling, or extended discussion. If you feel that this question can be improved and possibly reopened, visit the help center for guidance.
Closed 9 years ago.
My team is working on a new site which should be cross browser compatible (IE 8+, Chrome, Safari, Firefox, Opera) and cross platform compatible (Desktop, Tablet, Smartphone).
We've been looking at a lot of the new methodologies for achieving this such as HTML 5, responsive design, using JS libraries that abstract a lot of the browser mess away from the user but since the browsers and even MVVM, but the one major issue that I've been facing is how fast the browsers are changing. With both Chrome and Firefox using a model of continuous delivery it becomes harder and harder to test. And from the looks of it other sites have the same problem (it seems like it's hit or miss these days as to whether a site will work in a particular browser)
What suggestions do you or your team have for testing new browsers as they come out?
What things do you do during development to decrease possibility of having code break when a browser update comes out?
And how do you decide when you will or will drop support for a browser version?
NOTE: I've altered the order of your questions to leave the longer answer at the end.
Your questions
What suggestions do you or your team have for testing new browsers as they come out?
Actually, as you said, Chrome and Firefox are continuosly delivering so it eases the process. The last version you have is mostly always the version the user has.
For any other browser (and Chrome and Firefox old versions) just select a version of each and act as a "high pass filter", testing any version up from the one you selected.
How do you decide when you will or will drop support for a browser version?
Take a look at the statistics of browser use. There are many resources such as statcounter, w3counter, w3cschools, or wikimedia. If possible, add an analytics tracker to your page and you will have data about what devices, platforms, browsers, and versions of them the visitors use to access the site.
What things do you do during development to decrease possibility of having code break when a browser update comes out?
The key is to use a well defined methodology, based on the existing standards. Continue reading for a personal recommendation.
Workflow to ease cross-browsing
Step 1: Bootstrapping
At first decide: Graceful degredation versus progressive enhancement. Both are valid techniques, but makes sense using the first to fix existing projects and the second for newly created projects.
Then select libraries to avoid typing existing code, focusing on the 3 languages: JavaScript, CSS, and HTML. HTML5 (+CSS3) is the better choice today but support for older browsers must be provided. The following libraries ease supporting them:
modernizr for feature detection and conditional loading of js or css.
jQuery for ajax and dom related tasks.
normalize.css for normalize default browser styles, rather than just "resetting" them.
Notice that all of the js libraries listed above allow custom builds, an important thing when performance matters.
Html5 Boilerplate provides a strong template from which start the layout. It includes modernizr, jQuery, and normalize.css. Its github repository is a good resource to learn a lot about cross-browsing techniques. This article on its wiki has a nice set of links to start learning.
Step 2: Do the work
Designs should be mobile-first and responsive. This article on html5rocks introduces well why and how.
While "doing the work":
Follow the w3c standards. Avoids using hacks, specially CSS hacks, when possible. Review often the HTML5 specification as it is pretty unstable.
Take care of ECMAScript 5 features when writing javascript. Rely on libraries to avoid code breaks caused by deficient browser implementations. Do not extend the DOM.
Automate tests when possible. Layout and specially layout polishing, including animations, are manually tested cause it's quicker but UI functionalities like form submision can be perfectly tested with automated tests.
Use tools to ease tasks. Chrome + devtools or Firefox + firebug are the very basic must-use, but there are a bunch of tools to ease cross-browsing tests, even automating those tests.
Annex: Tools and resources
Cross-browser testing
Browserstack is just awesome. Allows testing on all devices, platforms, browsers, and versions.
Browserling is an alternative to browserstack. It is developed and maintained by Peteris Krumins and James Halliday, both recognized members of the node.js community and well-known developers. They also published a tool to automate the process called testling-ci, but this is only relevant if using node.js on the back-end.
modern.ie provides tools to ease testing on internet explorer. Developed by Microsoft, the site provides live testing through browserstack and downloadable virtual machine images with pre installed software.
adaptability testing for "responsive design"
respon.si is an online tool meant to test the visually appearance of layouts. It allows selecting a resolution so it's useful for responsive layouts testing. Notice that any other tool to select a resolution can easily do the same.
What suggestions do you or your team have for testing new browsers as they come out?
As part of our definition of done we support the following desktop browsers:
IE8+
Firefox 3.6
Firefox (latest)
Chrome (latest)
Safari 6
The support of the latest versions of Firefox/Chrome is fine because they both provide automatic updates, and so if anybody has a problem on an older version of the browser, it's out of our hands and they should update.
The majority of Firefox/Chrome testing can be done on our machines, but there are obviously discrepancies with how the different OS' handle fonts, and some quirks with native form elements that may or may not carry over to versions on Windows.
To test Firefox versions on OS X I use the "Install all Firefoxes" script that I created, to allow me to run multiple versions of Firefox side-by-side.
Our development team uses Ubuntu and Mac OS as their environments, so we have a dedicated machine with virtual machines for each version of IE, and Chrome/Firefox on Windows, and Safari 6 on OS X.
These virtual machines were setup using the images provided by modern.ie. We're remotely accessing the machine with the virtual machines on so that we don't need to break our workflow and go to another machine.
What things do you do during development to decrease possibility of having code break when a browser update comes out?
The obvious things are avoiding CSS hacks, and making sure that the HTML/CSS/JavaScript that is written meets our code standards, and our definition of done.
If we're using experimental CSS features, we ensure that we're providing vendor prefixed and lastly w3 definitions of properties:
-moz-foo: bar;
-ms-foo: bar;
-o-foo: bar;
-webkit-foo: bar;
foo: bar;
Obviously this introduces some technical debt, but if you're using a preprocessor for your CSS, the overhead of this can be reduced.
We keep a separate stylesheet for IE and load it using conditional comments, so that we can fix problems in IE without affecting the integrity of the rest of the front-end for other browsers. There's a movement lately to moving this in to a shame.css though, which you can read about here: http://csswizardry.com/2013/04/shame-css/
And how do you decide when you will or will drop support for a browser version?
Google Analytics. Segmenting by browser type in Google Analytics is very useful. When usage for a particular browser drops below 10% it's a good time to stop developing new features for it.
You could do something as radical as throwing away all of your hacks/styles for that browser too, but for a smoother transition (and to encourage people to upgrade) it's better to simply stop developing for the older browser, and maybe conditionally display a message.
We dropped support for IE7 recently, and now visitors using IE7 will get a message telling them to upgrade, and they don't get any more fixes or additional features.
Mobile is a whole other kettle of fish, and if your site is completely responsive it's an extra layer of pain.
We've got a bunch of different size/version Android devices, a couple of old iPhones and an iPad kicking around the office that we use for testing the majority of mobile browsers on.
Obviously there are differences with screen size, DPI, browser version etc. The best you can do in this regard is cater for the most common case, and fix any issues as they surface.
design and develop with web standards. validate the web platform. all of the libs/solutions/etc., already listed are phenomenal and quite useful, but if you stick to standards, you'll find you need those libs less and less...at least for gracefully degrading user agent support. i find i use them more and more for progressive enhancement.
I never thought, I'd have to ask THIS question:
Is there a cross-browser way to play background MIDI files in a website without the need for a plugin?
(Now, please don't answer with "don't do it! it's annoying as hell!". It's for a kids-page and they love this stuff... And I will give the option to turn it on or off...)
I've found several approaches, each with their own problems:
<bgsound ...>-tag: IE only; must be careful not to use with other tag that IE might interpret also
<embed ...>-tag: Doesn't work in Chrome. Firefox requires a plugin, which it can't even install automatically...
<object ...>-tag: Chrome and Firefox require a plugin. Both suggest QuickTime, but neither installs it correctly...
<audio ...>-tag: Requires HTML 5; also not sure if it supports MIDI across browsers...
use a Java applet: Not available everywhere, and seems like overkill
Is there some way to make this work across browsers (minimum: current Firefox, Chrome, Safari, Opera, IE6+, Android, iOS) without requiring any plugins? Probably some clever mix of all the tags listed above?
I am really surprised that this seems to be so difficult. Is MIDI deprecated? Or background-audio in websites in general?
Why not HTML 5 Audio. After all it is supported in all latest browsers. ( Firefox/Opera can have problem with H.264 files ). And in case of Mobile browsers it should more better choice.
One option you may want to pursue is SoundManager2. An audio project based on JavaScript. I use it in many of my web projects and so far it seems goood. Requires a little bit of a learning curve, but the API is very rich and clear and how to install it and use it for various situations.
http://www.schillmania.com/projects/soundmanager2/
By building a JavaScript package for audio within your web site, you won't have to worry about additional plug-ins that front-end users may need to have for their machines. It works with old and new browsers, MIDI files, MP3's and more.
What are the current best recommended resources for cross-browser testing in the CSS3 jquery world? I am adopting html5 and have adopted parts of css3 into my latest web project. I have also changed over to jquery for my scripting needs. The best answer I have seen for testing this pre-dates html5 and css3 being ready for primetime.
I am testing in Windows Opera, Safari, 2 Firefoxes+firebug, IE+f12 and two Androids. I test myself for input=expected output in my javascript and browser-sizing and other rendering issues. I test my php generated code by rendering it in firefox, then validating. I am trying to test for user experience by having myself and other people use the site and reporting on their opinion of the design, usability and overall impressions of the site-in-action. I present them with views on different sized devices. I even am lucky to have a friend with color-blindness;-} The F12 solution is particularly dissatisfying when testing in IE and the css rendering consumes large chunks of my development time.
Is better emulation available? Has anyone found a better (i.e. fast, easy, efficient, scientific) way to test across platforms yet? I am hoping for a strategy like unit-testing to emerge in the online community so we can make our apps more stable and robust as they become more powerful and influential. I can't seem to find a way to tame this chaos!
Have you used Selenium?
http://seleniumhq.org/
It is a tool to automate browser testing. Some people at Google made it and open sourced it.
You may want to look at RIATest for cross-platform cross-browser testing of HTML5 applications.
It works on Windows and Mac, supported browsers are Firefox, IE and Chrome. Automated testing scripts written on one platform/browser can be run against all other supported platforms/browsers.
(Disclaimer: I am a RIATest team member).
Is there any single framework with which I can build a css3, html5 website that is compatible for all browsers including IE7 and later? Can http://html5boilerplate.com/ boilerplate help me in this?
You will never get the IE7 or IE8 rendering engine to achieve full compatibility with HTML5, CSS3, and other modern technologies. They are simply not capable of it.
However there are some hacks, tools and plugins which can get you part of the way.
Tools like Modernizr will help you by allowing you to detect which features are supported, to give your site a chance to work around it.
jQuery is a great library anyway, but is particularly good in this context because it abstracts a lot of browser differences away from the developer. Some things are easy in most browsers but a real pain in IE; jQuery takes a lot of that kind of stuff and makes it easy regardless.
Dean Edwards' IE7.js and Selectivzr are both Javascript libraries that give IE support for lots of the CSS selectors which were missing in older versions. This allows you to write your stylesheets without worrying so much about what IE supports. (IE7.js also fixes a number of IE's other glitches and missing features too)
CSS3Pie is a hack for IE that adds support for CSS border-radius, gradients and box-shadow.
There are in fact a whole load of hacks along these lines, all aimed at adding features to older versions of IE which it is missing. Modernizr's website has a big list of them here: https://github.com/Modernizr/Modernizr/wiki/HTML5-Cross-Browser-Polyfills
However there is one big caveat to all of this. Speed. IE<=8 is a slow browser. It has a slow Javascript engine. Virtually all of these hacks are javascript based. You might get away with running a few of them on any given site, but trying to use enough of them to give IE anything like full support for HTML5 and CSS3 will slow the browser down to the point of being unusable.
There is one other angle to approach this question though, and that's Google's Frame plugin for IE. This basically installs the entire Google Chrome browser engine into IE. The user is still running the IE shell, but the web page is rendered as it would be in an up-to-date version of Chrome.
This sounds great, but of course it isn't perfect. The main down-side of it is that the end user has to install it manually onto their computer, which means that you as a web developer have no real control over whether it's there or not. So its not something you can just add to your site and expect everything to magically work.
Finally, you may also be interested in CanIUse.com, which gives browser support tables for various features, allowing you to see at a glance what is and what isn't supported in various versions of each browser.
HTML 5 is more an umbrella term for a variety of technologies than any one thing or framework. Really, any implementation you do should degrade gracefully-- and that is what makes the site browser compatible. Detect what the browser supports, and then have alternatives that are legacy friendly.
Check out: http://accessites.org/site/2007/02/graceful-degradation-progressive-enhancement/
And: http://diveintohtml5.ep.io/detect.html
Good luck.
Modernizr is an open-source JavaScript library that helps you build the next generation of HTML5 and CSS3-powered websites.
MS is including it in the ASP.net MVC template projects
html5boilerplate won't help you.
As others have said html5 and css3 are very broad terms, there's no framework that implements all of the features provided by both of these. In fact they are not equally supported in browsers that say that they support them. And more, even W3C guys (who write the specs for html5 and css3) are not yet established on what features they will include or what steps will be required to implement them. You get the idea...
But.. if you narrow your requirements to some specific features, like video, audio tags, canvas drawing, nifty css effects, File API, or others, you CAN find frameworks that implement (or try to mimic) them with available technologies in cross-browser manner (more or less).
Is anyone writing applications specifically to take advantage of google chrome?
Are there any enterprise users who are considering using it as the standard browser?
Yes, I have started to pay very good attention to Google Chrome for my applications. Recent analytics show that between 6%-15% of my users are accessing my applications (varies between 6 to 15 in different applications) on Chrome. And, this number looks on an upward trend.
Thus, I can't really ignore it for testing right now.
As far as taking it as a standard goes, thats a long way off. I still have to test for IE6! :( Though, we have been planning to start using features like Gears (inbuilt in Chrome - downloadable elsewhere) once Chrome crosses the 25% mark. Thats when I believe that we will be looking at Chrome to be our preferred browser. I hope that we have Chrome 1.0+ by then! ;)
I switched to Chrome and haven't looked back except for the occasional site which doesn't work properly, forcing me to load it in Firefox. All my existing web applications work fine on it, and I'm using it for primary testing on my current development project.
I'm not actually targeting chrome, but I have added chrome to my browsers to test sites on. I've found some odd quirks in this product where some plugins cause the browser to hang, or run really slow in some environments, but they are still in beta in active development. But I definately now make sure sites I work on render well in chrome, as well as firefox, latest versions of IE, safari, Konquerer and opera. I usually check out how it looks on lynx as well, that helps me catch "un-alternated text" in images. Yeah, I know that isn't a word, but some people will understand what I'm saying.
Because chrome uses the webkit to render HTML, you can be assured if it works in safari, it'll work under chrome, however it's rendering engine isn't up to scratch quite yet. I think writing applications that take advantage of it is similar to writing iPhone applications, remember chrome is expected to be adopted by android to make it similar to iPhone. That way it pretty much takes advantage of all those iPhone apps.
Would I install it as the browser of choice? not yet - but i'll certainly work on valid web pages that will render across all browsers.
One of our major customers has outlawed Chrome because it installs on the C drive without asking. They deploy a standard image with a small C drive and large D drive so they can easily re-clone the system part of the image on C without destroying the client's personal files on D. Most software allows you to choose the install directory. Anything that violates this is disallowed, and they're a big enough company to have some weight with most vendors.
We have enough headaches trying to support
Firefox
Two versions of IE which have their own iffy bugs
Safari
I'm not sure why we continue to support Safari. Most of our users (corporate) use IE6 or IE7. We try to make sure that things work in both of those.
Maybe not for programming purposes but Chrome w/ Google Reader makes for the most powerful RSS reader. Can handle up to 1500 feeds w/ performance still ok, managing subscriptions still functioning.
I'm using it on my work machine, but that's about it. It's been stable for me, and I like the barebones UI. I'll still switch to Firefox for the web developer extensions however.
I'm liking some of GoogleChrome- the Start page with your 9 most recent is the winner for me. The interface takes a little getting used to, but the speed is impressive, especially with Gmail.
However, it glitches with Java, which rules it out for serious work at the moment. I use FireFox mostly and have Chrome for the "other" websites at work.
I'm considering using GWT on an intranet project and considering suggesting to the users that use Chrome to take advantage of the enhanced Javascript performance. Any AJAX-heavy app would be a great candidate to target Chrome.
At my company, we're not targeting it, but we're definitely paying attention to it. My boss is using it as his primary browser, and I have implemented browser detection for it in our scripts in case we ever to need to target it for some reason.
Chrome has the .png opacity bug where the transparent parts of the .png are a solid color if you try to transition the opacity from 0 to 1. In IE7 the opaque parts are black, and in Chrome, they are white. Today, I decided to go ahead and account for this bug in my JavaScript. I don't really test sites on Chrome that often, but I am actually using it for almost all of my browsing.
I will target Chrome as soon as a stable Linux and OSX client is available.
Targeting Chrome/Chromium right now, I think is like targeting Konqueror web browser. It will get popular, but you should wait to a more stable beta, and/or some Linux and OS X client.
My website statistics shows 3.xx % visitors using Chrome which arrived just few weeks back. And Opera is only 4.xx % which has been around for several years.
Easily you can see that rate at which Chrome is picking up.
You can see how easily Google takes over all areas of your computing world and personal world too.
Since Chrome uses Webkit, it has the same rendering engine and DOM support as Safari (not necessarily the same revision of Webkit though). By testing in Safari, you can generally get by without worrying about Chrome. Any differences you find are probably just bugs that you should file on instead of work around.
However, because Chrome uses a different JS engine, there may be a few incompatibilities with Safari. So, if you're doing anything with JS, you might as well fire up Chrome and see if there's anything obviously wrong.
Generally though, you don't target browsers, you target rendering engines (with their associated DOM support and JS engines).
I am using Google Chrome, so far all the web apps I have work fine in it with no modifications.
No.
Why help Google further build an evil empire? In this particular case it is so obvious that they do not care about users but only obsessed with gathering usage info.
It's not any major player yet