Hacker News Reader

The 'fixes' aka hacks aren't that difficult.

The problem is it makes everything so brittle and people tend to do things the shitty route just to support IE.

Maybe it was the last changes to fix IE6...but I actually looked up browser percentages and considered dropping support.

IE + hacks, with respect to technical debt, is like one of those credit cards that offer 0% interest and free miles for the first 6 months after which it switches to a non-premium card with a 37% APR.

I estimate that supporting IE6 has added more than 30% of the interface development costs to the complex projects I have worked on that use a combination of sophisticated CSS and JS to get the job done. This is a HUGE cost for start-ups and for clients to swallow, once the true cost is explained to them.

Just supporting the way various sub-versions of IE6 inconsistently handle cookies is a nightmare, and in many cases lead to unsolvable issues. This costed a client of mine real $$$$ to discover, and less than 20% of their audience used IE6. Less than 15 people actually complained about the issue.

IE 7 isn't quite as bad, but it still adds a measure of time to front-end development efforts. I would estimate it in the range of 10-20%, depending on how complex the interface is.

On two recent projects, we dropped IE6 support, with virtually no push-back from a general-use audience. IE6 is almost 8 years old. Most people don't keep their cars that long, and can understand that an old crusty browser is not going to perform as well.

The only place where I would be hesitant to drop IE6 support is on marketing sites (where you need to entice someone to a call-to-action within the first 30sec-1min) and on e-commerce sites, where every dollar counts.

30% cost to support 70% of your userbase seems like a win to me.

This is only an issue for developers who target Firefox initially. That's the wrong way to go about web development. Run IE7 on your dev box and build everything against it. That's what your users will have, so that should be your first priority. Step two is making it work for FF.

The added benefit to that approach is that you'll find less cross platform issues. Going from IE -> Firefox is a lot easier than the other way around.

Are you kidding? Developing for a crappy browser first is the perfect recipe for an unmaintainable mess of code, since you will be relying on hacks to get IE to do render correctly, then adding hacks again for every browser that does the right thing.

First develop for a standards-compliant browser, then add (conditional) hacks to support IE. The other way around is just silly.

I think we posted this at almost the exact same time. Awesome.

I disagree. IE is the moving target. If you want maintainable code, it is best to build to standards, then hack to IE with conditionals.

IE has had the same box model since 1995. It is, and always has been the standard.

If you recall, Netscape used that same box model until version 6. So if you'd like to look for a point where standards were broken, look to W3C and Mozilla. It's certainly not Microsoft's fault that the "standards" organization decided to change the fundamental way things are rendered in a browser.

Not to mention that IE's box-model (in quirks-mode) is actually sane compared to the W3C standard.

see : http://books.google.com/books?id=ZCPWYFoWaMIC&pg=PA289&#...

Just because it's older doesn't mean that it should be the standard.

"Standards organizations" are there for a reason, so that issues like this wouldn't be so egregrious. Microsoft has had a while to fix it, and 8 years later far too many people are still using an old deprecated browser.

But IE6 isn't 70% of your userbase. Maybe 70% of your userbase is IE but only maybe 20-25% of that is IE6 so spending 30% more time/resources on IE6 is what I would call a major fail. But hey, that's just me.

And have you ever tried to develop complex ui sites on IE? I'm sorry...it just sucks and there is no good way around that. And I'm someone who knows IE6/7 quirks pretty good. Their script debuggers are utter crap and their developer toolbar is the most ancient thing out there. At least in IE8 they are copying Firebug...at that point it might not suck as much but who knows until I use it a while.

Yes. I developed the most complex UI site I've ever seen, Twiddla, on IE first followed by Firefox.

I never understood all the IE hate, to be honest. Until Chrome came out, I used IE7 as my daily browser. It's just better than Firefox in my opinion.

Voted up for developing Twiddla for IE.

I share your sentiment.. I develop against IE first then test it against the "good" browsers... which rarely screw it up. IE sucks for script debugging.. I mean EPIC suck. You're not even told what line of what file is screwed up (IE seems to concatenate everything into 1 file when giving the error report), but I've learned to rely on myself not to make common scripting mistakes, and created a cross-browser "console" object to aid in debugging.

Safari has the worst script debugging on the planet. It essentially tells you whether or not there was an error. Oh, wait... and how many errors, total. Thanks, Safari!

Get VS.NET, and suddenly IE has the best script debugging around. Firebug rocks and all, but VS.NET will drop you into a breakpoint will a full call stack, watch, step thru, etc. It's a proper debugger.

And yeah, it takes about the same level of effort to develop against IE as it does against FF. The advantage is that your stuff generally then works right in Firefox first try.

Um, Safari has a full Javascript debugger built in:

http://developer.apple.com/internet/safari/faq.html#anchor14 (among other google results)

Amazing.. I can do console.log(). And if there's an error (which the link provided conveniently has), I get to see the entire file which gave the error. So helpful... I now know which file to sprinkle console.log()s in. Furthermore, it doesn't work at all for dynamically loaded JS.

But, then again, there apparently were several Google results.

And somehow, you managed to cause a bug in Twiddla I have never seen elsewhere. When you make the window smaller, some of your content actually goes off the edge:

http://dustincurtis.com/screenshots/Team_WhiteBoarding_with_...

And it's not just bugs, but lack of features. For example, I have requirements for simple 2D graphics with client-side interactivity. If we only had to support Firefox/Safari/Opera/Chrome, we could use SVG, but for users with IE (which has neither SVG or Canvas), that means Flash. Doing simple graphics with Flash is well over 10x more work than just splatting some XML inline into your page (which I've also done for personal side projects).

Consider, for example, that Flash/IE is a completely different plugin than Flash/all-other-browsers. We've definitely had Flash compatibility bugs between major versions, between minor versions, and even between the "same version" on different browsers. I can think of a few that took many of us more than 30 minutes just to trace through, much less come up with a workaround. And all of this only because IE (even IE8, AFAICT) doesn't natively support any kind of client-side graphics.

I don't know where the 30% overhead estimate for IE came from -- maybe a simple site? That sounds really, really low to me. I would put it at least 100% overhead for the kinds of things I do.

You're mistaken. IE has had SVG support since IE5, long before Safari and Firefox introduced Canvas. And the nice kids at Google have built a little library that lets you emulate most Canvas functionality via SVG:

http://sourceforge.net/projects/iecanvas

(At first I thought this was a joke, but it's up to +4.)

I think you're mistaken. I tried IE6 and IE7 just now, and they don't do SVG (though it nicely offers to save the file).

I think you're thinking of something called "VML". IE does have that, but it's not the same as SVG, not as full-featured as SVG, and nobody else supports it. Maybe that would solve some of my problems, but I'd never even heard of it before -- though I'm sure checking it out now!

I think the URL you meant to post was <http://code.google.com/p/explorercanvas/>. I looked at the source, and it uses VML, not SVG.

Thanks for that. I'll update my acronym from here on out.

"IE does V" is pretty much all I was trying to convey. Good luck!

I agree totally. We decided a while back that we wouldn't attempt to support IE6 - life is too short. We do support IE7, but still spend a lot of time on fixes and workarounds for minor incompatibilities. If our stuff works on Firefox, it generally works on Safari too, with no extra effort, but an oft-repeated phrase in our company is an exasperated "f*^&&ing IE!"