ansaurus

Question

What do you wish you knew before you spent hours trying to fix a bug in IE6

Answer 1

+17 A:

I wish I'd known that I'd be spending hours trying to fix a bug in IE 6 over and over again for years on end and I really would have been happier delivering pizza.

eyelidlessness 2008-10-28 05:06:43

haha. Well everytime it happens, it only happens once. A lot of web developers are a walking index of IE6 bugs. I wonder how much money has been lost by web companies working to fix problems in IE6.

alex 2008-10-28 05:10:04

I don't know but there should be a class action suit.

eyelidlessness 2008-10-28 05:13:44

Imagine what life will be like when IE8 is out, and it's as standards compliant as they claim. Add a few years and IE6 *might* be all but gone.. IE7 is slightly less of a pain to develop for, with CSS anyway.

alex 2008-10-28 05:15:05

@alex: The next Windows Mobile is going to include the IE 6 engine. So much for it going away anytime soon.

eyelidlessness 2008-10-28 05:27:26

Also, to be fair, while IE 8 is fixing a lot of the obvious CSS problems, they aren't doing much to fix the DOM, Chris Wilson pretty much admitted as much: http://ajaxian.com/archives/ie-8-and-the-user-features#comment-266339

eyelidlessness 2008-10-28 05:30:34

@eyelidlessness: Are you serious? What a pain in the ass that will be.What is wrong with IE8's DOM? The javascript caching issue?

alex 2008-10-28 05:55:41

Almost everything that's wrong with IE 7's DOM, which is everything that's wrong with IE 6's DOM. Which is huge swaths of the spec, including everything from an incorrect implementation of Range to a complete absence of Mutation events to a completely broken event system itself.

eyelidlessness 2008-10-28 06:00:37

ahh. Can jQuery abstract this away? I have fell in love with this framework, although I feel I shouldn't rely on it.

alex 2008-10-28 06:06:14

jQuery doesn't touch Range and cannot (without lots of overhead or requiring you to use only jQuery DOM manipulation methods) simulate Mutation events; it abstracts the event model except it still can't support event capturing. There are other problems but those have been my biggest headaches really

eyelidlessness 2008-10-28 06:09:07

So what do these IE shortcomings mean to me as a developer? I'm sorry I'm not as knowledgeable about Javascript. You seem to know your stuff!

alex 2008-10-28 06:12:44

Oh, I forgot this one. At least through IE 7, Trident doesn't maintain its internal DOM structure as a tree, meaning you can actually end up with nonsense like a single DOM node spanning two adjacent parent nodes. You'd think innerHTML is the only way to do this, but it can happen with MSHTML too.

eyelidlessness 2008-10-28 06:13:13

The Range problems make working with text selections (especially in richtext) infuriatingly painful. No Mutation events means polling for DOM changes is expensive even though the browser must do it internally and could expose it. The fallout of the event model isn't as huge but can cause problems.

eyelidlessness 2008-10-28 06:14:59

(This didn't fit in the last comment.) Basically, it makes using the web as a serious application platform really difficult.

eyelidlessness 2008-10-28 06:16:02

I see now. If it's internally done, I would imagine it wouldn't be too difficult to provide events for Javascript to hook into? I assume Gecko/Webkit/Presto do this?Why can't a rogue MS employee just provide a forced update of Firefox 3 :P

alex 2008-10-28 06:21:55

Apparently there are no rogue MS employees, only XUL. Wait, that's the wrong line. I asked Chris Wilson (I think in the link I posted above) why they wouldn't simply embed Webkit in standards mode, and he informed me that this would break ActiveX (cont'd)...

eyelidlessness 2008-10-28 06:28:32

So I asked, why not include ActiveX as a browser plugin and everyone's happy? "ActiveX can’t be done by another plugin" Orly? Then a few weeks later, Google releases a beta of Chrome which does *exactly that*.

eyelidlessness 2008-10-28 06:30:03

And to answer your question, Gecko 1.9 (FF3) and at least the current release of Webkit support (most of) those events. Presto doesn't. But the point is that the engine *must* internally handle the events as specified, or reflow/redraw with CSS/DOM manipulation would never work.

eyelidlessness 2008-10-28 06:33:57

haha. Sounds like Microsoft! I didn't have any issues until I became a web developer with Microsoft. Oh well. Who knows they might make a last minute implementation.

alex 2008-10-28 06:42:00

Just to be clear, the Chris Wilson I mentioned earlier, who pretty much admitted the DOM is getting the shaft in IE 8, is the "Platform Architect of the Internet Explorer Platform team at Microsoft" (from his about page).

eyelidlessness 2008-10-28 06:48:23

They should do what Google did... throw away Trident and incorporate Gecko and as you said use the ActiveX Plugin. And they can make their stupid little GUI. And everyone would be happy :)

alex 2008-10-28 06:53:33

Sorry that was meant to be Webkit. But Gecko would work equally as good.

alex 2008-10-28 06:54:37

Webkit would certainly work better for their purposes. But could you imagine Microsoft using an open-source Apple project in their most popular software?

eyelidlessness 2008-10-28 07:02:06

Haha. Well it was originally from Konquerer wasn't it? Maybe they can grab it in it's Konquerer phase and *somehow* update it without making it the next Trident. They incorporated jQuery into their next .NET didn't they? I think they are beginning to get a far off scent of the coffee

alex 2008-10-28 23:20:07

To be clear, KHTML was replaced with Webkit in Konquerer.

eyelidlessness 2008-10-29 03:08:26

I think they'd find using Webkit unpalatable. "Hey PC, what's that new browser you're using?" But I would LOVE to see them use Gecko. And they wouldn't have to lose as much face.

Nathan Long 2008-11-11 16:28:07

Oh, let's be clear. The IE team has much more enmity with Mozilla than Microsoft has with Apple. I'm sure that Apple would rather announce that kind of cooperation as a partnership than as a sucker punch. I'm sure you've heard that Ballmer mentioned Webkit might be in IE's future?

eyelidlessness 2008-11-14 06:35:03

I have heard that in a conference in Australia. But do you actually think they'd do it? Apple and Microsoft playing nice? perhaps...btw, I delivered pizzas, and it was almost as stressful as css for IE6 (almost!)

alex 2008-12-04 01:12:05

Answer 2

+5 A:

Transparent PNG should have been supported...

VonC 2008-10-28 05:08:52

And the javascript/propriety hack is very ugly.. Why can't they at least add a update in a service pack to support 24bit PNG's alpha channels.

alex 2008-10-28 05:11:25

Single color transparent png's are supported though. I used that for a website with lots of 4 bit images and single color transparency, for they were usually smaller than their gif counterpart.

Martin Kool 2008-10-28 07:51:36

Yeah but alpha transparencies are way cooler!!!Single colour (I'm Australian :P) transparencies in PNG's are the same as GIF aren't they?

alex 2008-10-28 23:22:43

Answer 3

+2 A:

Probably that SELECT elements dont render with the correct z-order.

For example, if you have a floating DIV with a higher z-order overlayed on top of a select - the SELECT element will still render on top of the DIV.

Infuriating.

Adam 2008-10-28 05:34:32

I guess the only way to fix this is with IFRAMEs? yuck!IE6 must just paint that form control directly over everything

alex 2008-10-28 05:56:35

I've written code that will set the style to display:none for select elements which are positioned underneath the floating div. Kind of a hack, and you can see it disappear when it's only partially covered. But way better than the select covering up your floating menus.

Kibbee 2009-12-18 02:12:51

Answer 4

+1 A:

http://www.pushuptheweb.com/ or one of many other sites pushing to drop IE6 support. I'm aware that rendering web pages correctly for IE6 visitors is the core of many web businesses, but sooner or later something has to give. I just dont get why MS doesn't force an update >.<

Karan 2008-10-28 05:42:25

Answer 5

+7 A:

Number one thing: hasLayout If I had known about this from the start most of my worries would've been solved. I even regard it as a worse problem than IE6's stupid box model.

Joshua 2008-10-28 06:18:35

Answer 6

+17 A:

I wish I knew about quirksmode.org. The compatibility tables, bug reports, javascript examples are all very useful.

cofiem 2008-10-28 06:53:33

link goes to quirksmode.org but text says quriksmode.org.

Ray 2008-12-01 02:08:50

@Ray: that's an IE bug for sure.

eyelidlessness 2008-12-11 01:02:35

@eyelidlessness, I call that an IR bug

Jimmy 2008-12-11 01:13:34

Answer 7

+6 A:

I wish I'd known about

the conditional comments to include stylesheets just for IE
xhtml headers to make IE render in compliant mode
the box model problem, so that I knew what to put in my IE stylesheet

After I learned about these things, I haven't really used a lot of time fixing problems in IE6

.Hauge

Hauge 2008-10-28 10:01:32

You don't need any "xhtml headers" (IE does not support XHTML *at all*). Standards mode is triggered by certain DOCTYPEs, including HTML4 Strict DOCTYPE. Make sure DOCTYPE you use has URL.

porneL 2008-11-19 23:50:20

Answer 8

+6 A:

I wish I'd known about Position is Everything, but specifically the peekaboo bug has always got me.

Ross 2008-10-28 10:12:09

ForbiddenYou don't have permission to access /explorer/ on this server.

Chris Porter 2008-10-28 14:08:39

I think this is the link you intended: http://www.positioniseverything.net/explorer.html

Chris Porter 2008-10-28 14:09:12

Fixed the first link.

Quinn Taylor 2009-08-24 17:53:02

Answer 9

+2 A:

I wish I knew that my code didn't validate. Or that an XML declaration in XHTML puts IE into quirks mode.

roryf 2008-10-28 12:32:52

Answer 10

+6 A:

That if your really anal and spend shit loads of time on it that it pays off and you then become an IE guru fearing nothing but the sad self you have become...

Although it is nice to impress people with your amazing IE bug fixing abilities...

Most IE bugs can be avoided by using different (normally better) CSS methods and super clean logical xhtml
Always clear your floats with overflow hidden (or just hasLayout for IE)
Understand what hasLayout is (basically a css porperty that gives IE a kick in the ass)
When you start devving sites check IE6 all the time, untill your a pro ;¬P

Unfortunately, and I do train a small team of 6 developers, experience is one of the only things that really helps with these problems, stay calm, do good research in google and post your problem to a community if you really are stuck with a good demo of the problem.

Nice links >

http://css-tricks.com/ie-css-bugs-thatll-get-you-every-time/ (although I recommend the PNG fix by Bob Osola /-0)
http://www.gtalbot.org/BrowserBugsSection/ (amazing and funny... great!)

Good luck!

2008-10-28 14:00:36

Using overflow:hidden to clear floats is handy, but it won't work in all situations. Sometimes you need something (say, a graphic or rollover) to "spill out" of the div it's in, but overflow:hidden will clip it off

Andy Ford 2008-10-28 18:52:06

Answer 11

+10 A:

Don't code for IE6 first. That's the path to madness.

Ben Scheirman 2008-10-28 14:02:45

Amen. I develop everything primarily using Firefox 3 (and IE7 using IE Tab) as a test bed, then once I get everything working and looking good, I start trying to make it look the same in IE6 (and to a lesser extent FF2) without breaking it. Doing it the other way really sucks.

CMPalmer 2008-10-28 14:17:35

I agree, Firefox seems to be the best choice for developers. Develop for Firefox 3, then make compatible with other browsers. This seems to be the best way to keep your sanity.

Liam 2008-10-28 14:34:48

You might want to throw in Webkit (Chrome or Safari) and Opera in there for testing. Standing on the "FireFox is the WC3 standard" high horse can lead to a rather painful fall. FireFox has a lot of CSS rendering bugs, especially around UI controls ('input type=button' ignores the WC3 box model for example, which can cause alignment problems with other controls when viewing in standards compliant browsers). No browser currently works entirely as the WC3 standard dictates (Don't get me started on Opera's butchering of the CSS2 printing attributes) and I doubt they ever will.

David 2009-08-24 18:16:32

Ben Scheirman 2009-08-24 20:29:22

Answer 12

+2 A:

The main problem we have had is with scheduling enough "fix in IE6" time. That and the designers' tendency to come up with stuff that is easy to do in Flash and tricky to achieve in CSS has cost us many days of extra work.

pdc 2008-10-28 14:26:43

Answer 13

+1 A:

That even with all the PNG hacks in the world, if you're using PNGs as backgrounds in your divs (to make, for example, a panel with rounded borders), you're headed for a world of pain. (Links and other elements not being clickable, IE6 crashing in earlier versions of the png dll, etc.)

In summary: don't use transparent background PNGs if you want it to work in IE6.

Dave 2008-10-29 10:10:35

Answer 14

+1 A:

in addition to other IE bug lists that Ross and Loque gave above (gtalbot.com, what a beautiful site!

http://css-discuss.incutio.com/?page=InternetExplorerWinBugs

http://webbugtrack.blogspot.com/search/label/IE7

http://www.enhanceie.com/ie/bugs.asp

http://channel9.msdn.com/wiki/internetexplorerprogrammingbugs/ (like gtalbot, it'll take you hours to scroll thru this

and, hey, IE 8 beta 2

http://css-class.com/test/bugs/ie/ie-bugs.htm

finally, the IE voodoo doll!

http://www.flickr.com/photos/chisa/1349759901/

Gene T 2008-11-01 15:59:25

Answer 15

+1 A:

The importance of DOCTYPES in IE, and

Web Bug Track

scunliffe 2008-11-10 23:50:08

Answer 16

+7 A:

I wish I'd known why Microsoft hated me so much.

nickf 2008-11-11 00:18:40

This made me laugh out loud.

Nathan Long 2008-11-11 16:30:05

If IE6 was a person - he would have gotten a very very ugly slow humiliating and painful death by me. I would enjoyed every second of it.

DMin 2010-06-13 11:40:38

Answer 17

+10 A:

I wish I'd known that many IE 6 error line numbers are off by one.

Robert J. Walker 2008-11-11 00:48:44

I didn't know that...

alex 2008-11-11 03:47:04

Well, you just learned something new! (Doesn't that deserve an upvote?)

Robert J. Walker 2008-11-11 23:34:24

Answer 18

+3 A:

the underscore trick

if you put an _ infront of the css atribute it only gets read by ie6.

ie. _border: 1px solid #000000; creates a border only in ie6

2008-11-11 01:02:29

Answer 19

+1 A:

I Wish I'd Known That Internet Explorer Is The New Netscape Navigator 4.7.x

Eli 2008-11-11 01:20:17

+1 for the unfortunate truth

scunliffe 2008-11-12 04:46:39

-1 for spelling of Navigator.

porneL 2008-11-19 23:52:29

So we've progressed from "crashes at the very sight of CSS or any other WC3 standard" to "Renders CSS content, except when it doesn't and adds extra space anywhere it feels like"?

David 2009-08-24 17:59:06

Answer 20

+1 A:

The rendering quirk which causes a small white space underneath an image in a table cell if there is any white space between the end of the img tag and the closing td tag.

This renders right:

<td><img src="myimage.jpg"></td>

This renders wrong:

<td>
<img src="myimage.jpg">
</td>

Kibbee 2008-12-11 01:06:02

whats the best solution to this?

Simon_Weaver 2009-01-21 09:29:23

line-height:0px;

s_hewitt 2009-12-16 06:29:04

Answer 21

+1 A:

1

hasLayout and this excellent explanation: http://www.satzansatz.de/cssd/onhavinglayout.html

I had a new lease on life once I'd wrapped my head around it! Ensuring that the containing element hasLayout fixes the vast majority of my problems.

2

jQuery!! Writing IE-specific CSS is bad enough, so I'm glad I rarely have to worry about JS anymore.

chipotle_warrior 2008-12-17 18:11:51

seriously, CSS is more than enough. Thank god(or John Resig) for jQuery.

DMin 2010-06-13 11:45:57

Answer 22

+1 A:

I wish I'd known about dev tools like Firebug Lite and HttpWatch that make debugging lots easier.

Annie 2009-12-16 06:26:23

ansaurus

tags:

views:

answers:

What do you wish you knew before you spent hours trying to fix a bug in IE6

1

2

related questions