Time to scrap IE 5 support? – (and general web browser support discussed)
With the advent and following mass adoption of Internet Explorer 7, I’ve been pondering what web browsers to ensure support in, and which one to finally ditch. I’ll explain my choices below, but while reading the post, something like TheCounter’s Browser Stats for February 2007 can be a good reference point.
IE 5
Let’s start with IE 5. IE 5.0 was released in 1999 and IE 5.5 first saw the light of day in mid-2000. That is almost seven years ago! As you might know, these versions are full of bugs, no support for the correct box model etc etc. I’ve been discussing this on and off with following web developers, and I think that, finally, it’s time to stop wasting our time on these versions. Therefore, I propose that we stop checking things in IE 5. Good riddance!
My general stance about web browser and various platform support
Naturally, I think that as a front-end web developer, the obvious way to go is to support as many web browser and platforms as possible. However, there’s one thing to write good and semantically well-written code, and a completely another to work through various web browser inconsistencies and bugs just to make something work like it should have done in the first place.
For me it’s hard to motivate, most of all to the customer paying for the web site, all the time (and consequently money) put into making things barely hold together in old and outdated web browsers, especially if the web browser/-s in question have a miniscule share of the web browser market.
Do your research
Of course this depends on what web site you’re building, and if it already exists or is a completely new one. When I’m about to work on a public web site with lots of visitors, I make my homework, reading the statistics of the web site in question to see what the user base has as their preferred web browsing tool (just a little over a year ago, the second most used web browser for a large Swedish web site was Netscape 4, with an astounding 4%…).
I don’t develop for “dead” web browsers
My definition of “dead” is a web browser that was released over five years ago, and its market share is unessential (e.g. IE 5), or if developing of the web browser in question was stopped over a number of years ago (e.g. IE 5 on Mac).
My general web browser and platform testing
Roger wrote recently about what he tests in in Browser testing CSS and JavaScript, and I can right away say that I’m far from that ambitious. My aim is to support the large web browsers, covering roughly 97 – 98% of the web browser market, and my experience is that if you waste too much time on marginal web browsers such as IE 5 (on PC or Mac), this leads to you rewriting your initially lean code just to make it all work for about five more end users.
The cost of this is that the web site gets full of redundant code, and maintenance is a bitch, because whatever you change will break something in web browser # 19, that you just missed to test this particular scenario in. So, except for the obvious ways to go with separation of content (HTML), presentation (CSS) and interaction (JavaScript), using semantic code in general, and making sure that the web site is built on a solid ground working without JavaScript, the number of web browsers and platforms I test in are limited.
It should work in a lot of other web browsers as well, but I just don’t take the time to test and make sure a 100% that it works perfectly in them. So, my testing generally consist of these web browser and platforms:
- Internet Explorer 6, PC
- Internet Explorer 7, PC
- Firefox (about 1.5+), PC
- Firefox (about 1.5+), Mac
- Safari (about 1.3+), Mac
- Opera, occasionally (and then 9+)
All versions of IE 5 are as good as dead to me, and since Firefox seems to be the default for most Linux uses (I can be very wrong here), I take for granted that it will work if it works in Firefox on PC and Mac. What I haven’t mentioned are cell phones, and no doubt they are becoming an increasingly interesting target market. Personally, as of now, I don’t do any testing in any, but have so far instead relied on semantic code and proper CSS.
I also think that the most important part of a web site is the content and that it is accessible; not the design, not the interaction. Therefore, I’d rather suggest serving web pages without CSS and JavaScript to less competent web browsers, than trying forever to find a workaround for something that obviously wasn’t meant to be.
So, that’s my take? Do you agree, or find me nonchalant?
You’re absolutely right regarding IE5. However, more and more I find myself building websites where it only requires a couple of tweaks to get it usable in IE5. As such, I don’t want to say I won’t support it by default. Instead, I’ll just take a look at it and try to determine the effort required to get it to a point where it’s acceptable.
My browser testing is about the same. I’ll stick to the latest versions of Firefox, Safari and Opera only. Their buggy behaviour is too inconsistent (and too hard to target individually) to try to fix it in older versions.
And call me lazy as well, but I am exactly the same with cellphone browsers. I rely on my intentions to write semantic markup with seperate presentation from behaviour from content produces a usable document for such browsers.
On a side note, I do try to ensure that everything works 100% right in it. It’s doable for IE6, IE7, Firefox and Opera. Safari is a bit of a blur to me π
I generally build a first version using FF and then test in IE 5.5, 6, 7 Win, FF 1.5 and 2 Win, Opera’s latest version and Lynx and adjust where (if) needed.
I don’t have a Mac on stand-by for testing, nor a Linux machine. Most of my customers work the consumer market and so far neither of those OS’ have ever crossed the 1% boundary on the stats. Still, if I’m in the vicinity of one or the other, I quickly do some tests on whichever browser the machine has available to check that the essentials work.
Also for big big projects, I ensure that some befriended webdevelopers test the site on Linux and Max.
None of the projects I work on change so often that mobile browsing really becomes an issue, but still, if it works in Lynx, it should be fine for most mobile phones AFIK.
You might be interested in the browsers that the BBC develops for. The guidelines document is here.
IE5 is a level 1 supported browser meaning all content and functionality must work, and variations in presentation must be minimised. I've taken a quick look at the browser stats for the bbc (unfortunately it's an internal only document so I don't think I'd be allowed to publish it), but IE5 and 5.5 taken together have a larger share on bbc.co.uk than Safari (stats taken during November).
I think there's an assumption amongst the web development community that no-one uses these old browsers, or that the numbers are so small as to be insignificant. You really need to take the statistics from your own site with a pinch of salt, most "normal" web users wouldn't visit your web site, and probably wouldn't know how to upgrade their web browser.
I do agree that we need to end the life of these ancient browsers, but three percent of users using IE5 is still relatively significant.
I want my sites to be usable in IE 5.5 (PC). They don't have to look pretty, but they should work. I don't care for IE5 Mac – surely nobody is using that anymore?!
I think it always depends on the project. There are projects that should not be slowed down by incompatibility with old (sometimes very old) browsers.
I’m like you, I trust that my clean markup and CSS which is not full of hacks still looks great on IE 5 except for the box model situation. In recent testing my IE 5 test have gone just fine so who cares about a little width problem.
It was great after looking at my companies analytics, we are an ecommerce company with about 18 sites. The stats for IE 5 were .5%, so yeah thats a great feeling to know its just about extinct.
Oh, to see IE5 vanish.
Generally, I try to make things work in IE5, but if there are some minor layout inconsistencies, I’ll let it go. IE5/mac is a different story. I’ll definitely hide the stylesheet from that browser unless its required for a project (which hasn’t happened in awhile)
Georges is right. It depends on your project. If you develop for an existing project you can analyze the logs. If only one user per month comes with browser X I wouldn't bother testing this browser (if my page is overall valid).
Apart from this I think everyone who uses IE5 in the year 2007 deserves pain (except he's a paying customer ;-))
Sounds good to me. If people are still using IE5, I am sure that they’re used to seeing a jacked-up looking version of the Internet. If that’s the case, and they still haven’t switched browsers, then there is not much we can do for them.
Let's get rid of IE5 π You can't keep on developing for (the most) downgraded browsers anymore, it's simply not cost effective, time consuming…
and 7 years, itch…. rings a bell π
I look at it pragmatically.
If the amount of users that visit the site using IE 5, or IE 5.5 doesn't justify the time of even thinking about it I scrap the notion.
In a "web purist" standpoint sure you should support any and all browsers down to IE 3.02 or whatnot but from a business stand point I rarely support IE5 or even IE 5.5
As far as worrying about different browser and Linux platform combinations, I'd say, from my experience, don't.
I develop using Linux–I just moved from Debian to Ubuntu Edgy Eft–and have 5 browsers installed; Firefox 2, Ephiphany (same as Firefox), the latest Opera, and Internet Explorer 6 and 5.5, under Wine (IE7 is in beta).
Comparing any of these browsers running on Windows XP, I don't see much difference in the way most Web sites are rendered, except for the problem of fonts.
Like Robert, I don't want to degrade my clean semantic markup with unnecessary hooks, or divs and classes, to control IE 6 and 7, let alone 5.5. And, I don't see anybody visiting my sites anymore on anything less than IE6.
I quit supporting 5.5 more than a year ago: it was too hard and too joyless.
My company has hundreds of clients whose sites use heavy CSS styling. In the past two years, we have not supported ie5 or ie5 Mac and we never received one complaint. Maybe a very small percentage of people saw a funky web page here or there – but it sure has made coding, testing and maintenance easier resulting in lower quotes and faster turn-around for our clients. It depends on the project but for small/medium volume sites, I don't see much of a business reason to support 1% of the market (this includes Opera).
I wish I could!
For personal sites, or (some) new designs, I usually scrap V5- browsers. But you wouldn't believe how updated are the computers accessing the intranet I'm working on!
Thanks all for very good feedback, and fair and balanced comments!
And I agree: it is about context. If it is for a web site where statistics show that the percentage of IE 5 users are just too high, naturally you have to cater for it. But if they aren't, or if it is for a new web site, I'm all for letting it go.
I never write CSS hacks, but have had to get used to"stating the obvious" to overcome inheritance problems and doing a few other things to overcome "haslayout" as well as other well known box model problems. Doing this to make IE6 work, often means that IE5.x mostly works so the overhead of supporting it is not always a problem.
I agree with your original line of suggesting that it is such a minority interest, that it is time to drop explicit support for IE5.x
Of course to play devils advocate, and appear to be a heretic in many people's eyes, I *could* also take the view of not supporting Mac browsers of any form, they equally being a small minority :0)
Agree about getting rid of support for IE 5.5 but what about the fact that IE 7.0 was released as a general update recently??? So how soon will it be that most users do not use IE 6 and we can scrap supporting that as well…
Rob,
Absolutely, about the devils' advocate part. However, since most Mac web browsers are better than stuff like IE, it's seldom a situation you will find yourself in.
Jermayn,
It will probably take some time, but one can always dream. Personally, I look forward to IE 8, since IE 7 is pretty bad too (although much better than previous versions).
Since the release of IE7, web development has become a lot easier … most clients are willing to abolish IE5x support. This enables us to do strict mode rendering in both IE6 and IE7, which makes life a LOT easier.
That said, all the usual stuff applies, progressive enhancements, develop for the users, not the user-agents, blah blah blah π
If you go through the trouble of learing semantic markup, read up on how to make the web more excitable, ect & then knowingly leave ie5.x with a messed-up/unusable layout, (IMO) you have been wasting your time & failed as a web developer.
If you are not going to support ie5.x you should hide the stylesheet (as robert said) from it, just as you should have done for NS4 in the past (MS have made it very easy with conditional comments).
I honestly reckon we can start scrapping support for IE 6.0, well at least start fazing it out. Does anyone have any stats on just how many use IE 7.0 and IE 6.0???
Jermayn Parker, wake up! π Haven't you noticed that link top of this article?
Thanks for your comments. And yeah, Jermayn, please see the article at the beginning of the post. π
Whoops I missed that link, thanks for the wake up
Jermayn,
No problem! π
I too look forward to the fast approaching day when we can sto supporting IE6 (2 years, maybe?).
In general, when determining what browsers to support, looking at a sites browser stats is only part of the picture. Also important are how much effort will it take to achieve at least partical, if not full support, and what do the trends suggest that the landscape will look like in the next few years. To that end, I think both Apple and Linux will only go up, and therefor, should be catered to.
As far as mobile devices goes, while they too are rising in market share, because the landscape is still changing rapidly, and because page delivery is often interfered with, the proper direction to take is much less clear. If you need to support mobile devices, Opera, from what I've read, has good support and an increasing market share. Beyond that, I don't know.
Robert, as far as what Linux users use, Firefox is clearly the most popular, but Konqueror shouldn't be discounted. Currently, Konqueror probably has the best support for CSS3 and was the first publicly released browser to pass Acid 2. Personally, I tend to use it almost as much as I use Firefox.
Mordechai,
Very good input. Thanks for your comment!
Interesting that IE5.* has a bigger market share than Safari – the same is true on my website, but not by any significant margin; IE5.* has 1.0% and Safari has 0.9%.
The good news is that IE6 is now down below 45% – rejoice, rejoice!
I'm using ajax,css and xml in design of my sites.
I think, that's a good idea to leave older version of ie.
because I can't imagine anyone use ie5 after years.
[…] for the majority of (mostly) static sites, this is an interesting suggestion: serve older browsers the HTML without any CSS or JavaScript at […]