Standards Ignored as Developers Target IEMonday July 8th, 2002Dave Hodder writes: "CNET News.com has a story entitled Sites bow to Microsoft's browser king covering how Mozilla and other browsers are locked out because of web builders not supporting proper standards. It mentions Netscape/Mozilla standards evangelism and has some good quotes from Mitchell Baker, including the conclusion: 'What we're seeing with Web sites that are viewable only with IE is the privatization of the Web.'" . I never realised how often I grabbed the keyboard to shorten URLs as I surfed around the web until I installed Diggler. Just two little quibbles, firstly please, please make it work with the Modern skin. Secondly how about making it possible to chop the CGI variables bits of URLs off too, eg change http : //www .somesite.com / app.cgi?chop=this&off=too down to http : //www .somesite.com / app.cgi Would make a nice little addon even better. Thanks for the hard work. They write, that most TechEvang-Problems are Microsoft's "Extensions". That's wrong, most of these pages depend on MSIEs awefull error correction, many would even fail in a MSIE6 running in Standards mode. The second parts are morons, who still hasn't learned, that document.all is depricated. Another point, that I see more and more is, that if you write real valid HTML and use CSS2 up to it's end, the frist browser that fails (except Netscape 4, which is simply outdated) is MSIE width Opera short behind, while getting konqueror 3 to get crazy is way more complicate, on Mozilla it's nearly impossible without making the page unusable for the others. Afaik there is not even one page in the TechEvang list using some of Microsoft's CSS-Extensions, which would break for other browsers (as writing-mode), more there are pages, that still doesn't know, that MSIE isn't the only one supporting iframes or CSS on Tables, they normally have never heared of Netscape 6 or Opera! I agree with TheK. The sites I see that getting mangled using Moz are most often those with cut'n'paste scripts that are years out of date. http://www.webstandards.org/buzz/archive/2002_07.html#a000056 Palladium shmalladium... Standards are great and all. But when it comes to advanced CSS layout, some of IE’s “incorrect” interpretation just makes more sense than what the W3C dictates. For example, you would think that a z-index of -1 would lower that element one level. But in Mozilla, it puts it below the <body> tag, thus making it invisible. I wasted an hour trying to diagnose this. Another idiosyncrasy involves negative margins for elements in table cells. IE accept negative margins on all sides. Mozilla and Opera, however, reject a negative margin-bottom. There is probably some hunky-dory W3C explanation for this, but IE’s implementation just makes more sense and saves people like me time and frustration trying to track down the problem. Are the standards really a "problem". It seems to me that being a designer, you should know what you are doing. I am just a web surfer, but it seems to me that if something is a standard for web coding, it is only a "problem" to those that just thumb their noses at it. Yes, you are only a web surfer. Just like you, many designers don't know what "z-index" is, never mind arcane technical specs like a negative z-index bringing an element below the viewport. I'd hardly call that thumbing my nose at it. If I were to "thumb my nose" I wouldn't care that Mozilla makes elements with negative z-index invisible, thus rendering them absolutely useless. Instead, I would just tack on a warning that says, "Best viewed with Internet Explorer because Mozilla screws up negative z-index's and it isn't worth the trouble to figure out why, or implement some arcane workaround." Like it or not, IE is the defacto standard. When testing, IE is the first (and sometimes only) priority of web developers. Discussions about webstandards tend to be rather religious and hence not very productive. Basically you have two camps: the "w3c==holy" camp and the more pragmatic "I don't care as long as it works in IE camp". I've converted my site to XHTML 1.0 strict and css and am a mozilla user. I've already encountered a few things that work differently in Mozilla and IE, even though my site is very simple. It seems that standards compliance doesn't give you much more than a pretty 'validated' icon on your page these days. I can well imagine that web developers are pissed off by the w3c standards since they can be quite complex and rarely work as specified on the primary deployment platform (ie). CSS seems like an overly complex, terribly limited standard (most word processors, even the older ones, offer far superior layout features). HTML is one big compromise of inadequate structural tags and bolted on, poorly designed layout features standardized after the fact (i.e. after the 4.x generation of browsers was deployed). The only standard that works more or less reliable across browsers is the HTML 3.x standard. It doesn't surprise me that there is so much misinterpretation of and confusion about the css and html standards and I am pessimistic this will ever change. Both are the result of design by comittee and that's not a process that delivers the most optimal solution. Dealing with implementation differences between browsers requires intimate knowledge of both the standards and their various implementations. Most web developers don't have the necessary training for this and will try to work around issues by copying code that works for them, incorporating so-called css hacks and ocassionally accepting that stuff won't work on some browsers. I'm sorry but this 'IE is a defacto standard' talk is always total crap. To be a defacto standard, IE would have to HAVE a standard, and that's simply something that it doesn't have. Most of the time things that IE does aren't clever or useful interpretations of the standards, they're just plain errors. The problem is that web developers have somehow got the impression that the way to write web pages is to type some random html into a text editor, and see how it looks, refine it, see how it looks, etc. Imagine if people programmed like that? What people should *actually* be able to do, is design their site, and type html (and css) into a text editor, and know how it will come out when looked at in the browser, without ever having to do so. Obviously it would be foolish not to check, but the process should only require that at the end for aesthetic decisions and to see whether you've made any typos. I've developed a few medium-sized websites, and I tend to do them by reading standards and references, and then typing. 99 times out of 100 mozilla does just what I think it should, and this makes life easy. IE, however, just has random weirdnesses that just cause confusion. My most recent example was where I had a four columned layout and one of the columns was a few pixels higher than the rest. It turned out that IE randomly did this because I had a second paragraph which wasn't a full line long. Who can say what they were thinking if this was intentional - I'm certainly inclined to seriously doubt they were thinking. So what am I saying in my long-winded way? Basically just the IE isn't a standard, it's a set of mistaken deviations from standards. Nobody would claim that there is an IE PNG standard - it's just an attempt at the PNG standard which doesn't fully work. I could easily imagine a situation where IE would be a defacto standard - where rather than reading top as the top it read it as the left, or whatever. If this were so, they could publish the specification, and you'd still be able to type and then view, rather than having to constantly work around idiocycrasies. As it is, it isn't any sort of a standard at all..... Yes, but Microsoft's deviations from the W3C spec act as a standard. Microsoft provides in-depth documentation of IE's extensions to the standard and deviations from the standard at http://msdn.microsoft.com/ . In addition, M$ has their own doctypes, which Mozilla.org lists at http://mozilla.org/docs/web-developer/quirks/doctypes.html . So while Microsoft doesn't actually publish a standard, they provide enough information to developers for them to use it as one. - MXN To go to the w3c website and read through the chapters of the HTML and CSS documentation explaining the basic ideas behind these two languages which should be enough to understand why something acts a specific way in standards compilant browsers. Or to search through the MSDN and read the thousands of notes that explain thousands of issues the different IE versions have with HTML? It doesn't take that long to learn HTML and CSS and write even very complex layouts with very litte markup, what is the greatest feature of CSS in my eyes. Since I realize my layouts with standards compilant CSS my pages are faster done and easier to manage. I never encountered a layout that's impossible to do using only standards CSS only difficulties when IE didn't understand these and doesn't offer an alternative. ...again The problems are _not_ (!!!!!!!!!) Microsoft's Extensions, they aren't used at all! The problems are Microsofts nowhere documented "error correction", where they sometimes more guess, what you want to display, then showing what is written there and the second problem is document.all, which some morons will use forever, I guess some won't even stop using it, if the MSIE will drop it. There are a quite a few things that full standards compliance opens up that crufty IE html/javascript/css just make impossible. I have seen (and done) advanced things in Moz that just would not work in IE. About the only thing that is really killing me at the moment is support for the adobe SVG plugin (which needs work from both sides). The w3c's recommendations have proven far more enduring than the multitudes of proprietary extensions that have wired into HTML by one browser or another. Not to mention more usable, sensible, futureproof etc... I'm sorry to say but that is a load of bullshit. Essentially you say that webdevelopers should abandon their tools and use text editors instead. I'm sorry but the real world is different from your imagination. I know how to use an editor and in fact my site complies with relevant standards. However, I'm not a web developer but a Ph. D. student with way to much time to look into this sort of things. A web developer typically either has no CS degree or not a relevant degree. He/she generally uses tools like dreamweaver or even frontpage and just wants the damn site to work in IE where most of the target audience is. Bare HTML/CSS is way too complex for them to handle properly and they need to rely on tools to keep them from shooting themselves in their feet. If you use more advanced features in such tools, you lose standards compliance. Newer tools are better in this respect but still far from perfect. Now if I were a webdeveloper using such tools I'd want all the latest stuff to make my sites look cool. As far as the w3c is concerned, they produce interesting standard <i>proposals</i> but the real standard is whatever the browsers support. That's what I call a defacto standard. Unfortunately the defacto standard is not the same as the w3c and its supporters would like to see. It is either some subset of HTML/CSS that works more or less consistently or whatever IE provides. You can make the moral judgement of IE being an inferior product and MS being evil. That doesn't change their marketshare however. Web developers have basically three choices: code to the limited subset of standards and accept that advanced features won't work, code multiple versions of a site or code for IE only. The latter option has the best tool support and allows for the most advanced designs. That makes it attractive. You mention consistency: there's nothing better then real standards. Browsers get better and more standards compilant from version to version. So if you code to W3C standards chances are high that it will work in future browsers. Now you tell something about the IE 'defacto standard'. This 'standard' changes from IE version to IE version. On this machine here I have 4 different Windows installations with 4 different IE versions to be able to test my sites on every of these and they behave very differently on some issues. So how can this be a 'standard' if it's not even consistent on different versions of the same product? In addition to the inconsistant IE rendering on Windows, just look at how much IE on the Mac behaves different than any version of IE on Windows. Maybe webdevelopers don't care about the future but need their sites to work today with the crappy browsers that completely dominate the market now. I'm not saying that IE is somehow better than the w3c standards. I'm just saying that both are crap and that developers develop for a mostly IE dominated world where w3c standards don't really mean much. Given such a context it is pretty understandable that web developers mostly ignore w3c standards and go for what actually works in their browsers and tools. Telling such developers that they are idiots, that they should abandon their tools and bow to the holy w3c is IMHO pretty unproductive. The way out is not being religious about the w3c standards but being pragmatic. Web developers need toolsupport for what they do, if these tools stay within the standards, a large part of the problem goes away. Dreamweaver for instance is on the right track with (some) support for XHTML and a more or less working validator. To be honest, I've seen better standards in a lamp factory! I think there is no very much problem with simple HTML and CSS because MSIE 4+ works more or less with HTML and CSS W3C standards. The problem is scripting the DOM (DHTML), MSIE 4,5 and 6 (included) document object model is far of compliance with W3C-DOM-2. MSIE 6 complies with W3C-DOM-1 but it lacks of standard event and CSS models (defined in level W3C-DOM-2). But MSIE (6) is converging with W3C (and Mozilla as a very good W3C browser), a good notice, compatibility problems are not so hard as Navigator 4.x against MSIE 4 era. To bridge the gap: not use MSIE specific stuff, and use cross-browser libraries. I recomend: - Yasd's X-Objects http://www.yasd.com/dynatech/xobjsintro.htm - CBE http://www.cross-browser.com/ - My own's XPDOM, a cross-platform W3C-DOM-2 compatibility layer to Web applications. http://xpdom.sourceforge.net Enjoy I think there is no very much problem with simple HTML and CSS because MSIE 4+ works more or less with HTML and CSS W3C standards. The problem is scripting the DOM (DHTML), MSIE 4,5 and 6 (included) document object model is far of compliance with W3C-DOM-2. MSIE 6 complies with W3C-DOM-1 but it lacks of standard event and CSS models (defined in level W3C-DOM-2). But MSIE (6) is converging with W3C (and Mozilla as a very good W3C browser), a good notice, compatibility problems are not so hard as Navigator 4.x against MSIE 4 era. To bridge the gap: not use MSIE specific stuff, and use cross-browser libraries. I recomend: - Yasd's X-Objects http://www.yasd.com/dynatech/xobjsintro.htm - CBE http://www.cross-browser.com/ - My own's XPDOM, a cross-platform W3C-DOM-2 compatibility layer to Web applications. http://xpdom.sourceforge.net Enjoy standards may be important and all, but it's unlikely that many people will switch to mozilla and/or netscape over standards compliance. mozilla has plenty of other good selling points that make it superior to the micro$oft broswer, and that's what should be emphasized more so than standards evangelism. and if ditching standards in parts is what's necessary to gain recognition and general usage, then mozilla may just have to incorporate some of those horribly frightening m$ie nonstandard elements. The idea is that standards should encompass the best and most complete feature set for a language. If someone proposes something good (and futureproof) then it will probably make it into the w3c recommendation. If IE's nonstandard elements are the best solution to a problem, they will be incorporated into the standard. IFRAME is a good example, which I think was included in IE 3, and added to the HTML4 spec. MARQUEE on the other hand is a step backwards from where HTML was heading so it was never accepted. The point of standards is that they are open - they are freely available to evreyone and everyone has a say in it. Do you hear Microsoft asking for submissions from users about their HTML implementation? Remember that w3c recommendations are contributed to by industry figures including M$, it's not as though they have no say in it. If it is GOOD, it should be standard, and eventually will be!! Moz should not follow IE around - that would be asking to be screwed over by M$. Moz should go with the group concensus rather than with who can yell the loudest. This is the same company that can't even be bothered to update Frontpage to work with the latest version of Apache http://newsforge.com/newsforge/02/07/08/1358241.shtml?tid=2 . So remind me again, how commited are they to standards ? Frontpage, one of the more popular webauthoring tools, is a MS product. Now which "standards" do you expect it to support? Want more standards compliant pages? Make a good standards compliant HTML authoring tool available. Composer dosen't cut it for anything other than tossing togather a user's homepage, but the codebase might be a good starting point for a tool that a pro would want to use. Frontpage is not really a "popular" webauthoring tool as much as it is a "common" webauthoring tool. Most people do not choose to use Frontpage because they think it is great; they choose it because it is readily available. Often people who use Frontpage have never tried anything else and they may not even know the name of any other web tools. ..because the problem here really stems from lazy developers not desiring to do it the correct way. -Chris |