Comments on: Standards fail.

Standards fail.

jaduncan — Thu, 16 Oct 2008 07:26:12 -0800

Only 4.3% of the web validates. Opera have finished a scan and validation check of the net using their new MAMA spider and have got an extremely interesting dataset. Did you check your website today?

By: krinklyfig

krinklyfig — Thu, 16 Oct 2008 07:36:42 -0800

My websites always validate. I haven't had a paying web gig in years ...

By: DU

DU — Thu, 16 Oct 2008 07:36:50 -0800

MetaFilter: "look at my kittens" sites

By: smackfu

smackfu — Thu, 16 Oct 2008 07:39:58 -0800

Validation is overrated.

By: DU

DU — Thu, 16 Oct 2008 07:40:40 -0800

I'd like to add a major reason to his list of reasons so few pages validate: Confusion. The HTML/XML/CSS standards have gotten big. Worse, compliance is no guarantee of correct behavior in actual browsers. When I've made pages in the past, I've worked hard to get compliant. But many times I've just thrown up my hands and eventually done Something That Seems To Work. I'd like to see some stats on what the compliance rate is among the subset of standards that ARE implemented correctly across all browsers.

By: Thorzdad

Thorzdad — Thu, 16 Oct 2008 07:41:18 -0800

Simply adding any Google analytics or sitemaps code in an otherwise valid website will make it fail validation, depending on your stated doctype. I haven't time to read the complete report...Do they make any sort of account for this?

By: mandal

mandal — Thu, 16 Oct 2008 07:44:23 -0800

I wonder how many this little bugger accounts for: &

By: East Manitoba Regional Junior Kabaddi Champion '94

East Manitoba Regional Junior Kabaddi Champion '94 — Thu, 16 Oct 2008 07:56:22 -0800

obligatory

By: eustacescrubb

eustacescrubb — Thu, 16 Oct 2008 07:58:07 -0800

Either the whole point of having a "standard" is so that there is a baseline of functionality across the browsers or the whole point is to follow the rules Because They Are The Rules. Most of the people who build websites professionally can't pitch the second option to clients. There has to be a benefit to choosing the option that requires more work and has more restrictions on the final product -- so I find it's easy to pitch CSS-based layout/design, and most of the practices advocated by Zeldman and the like, but my clients don't care if a site validates -- they care that their customers can all see and use the site; they care about Google ranking; they care about functionality. If validation takes more time and effort and restricts what can be done design-or-function-wise but offers no benefit to the client's bottom line, then the client won't pay for it. So, these results aren't surprising at all.

By: Brandon Blatcher

Brandon Blatcher — Thu, 16 Oct 2008 07:59:47 -0800

One of my sites (PoliticalFilter.com) validates fine until links start getting added to it, then validation fails because the URLs constructed by other websites, which I have zero control over, aren't made to a certain standard. Yet the links work fine in that people can just click on the link and go to it. Who's wrong and why should I care if I can't control it and everything works fine anyway?

By: davemee

davemee — Thu, 16 Oct 2008 08:05:54 -0800

Thanks, EMRJKC94. I shall take a pro-active stance, express my dissatisfaction, and no longer interact with any site that fail vali

By: mandal

mandal — Thu, 16 Oct 2008 08:07:37 -0800

Dammit East Manitoba Regional Junior Kabaddi Champion '94, now there's recursive failure too.

By: effbot

effbot — Thu, 16 Oct 2008 08:11:37 -0800

the URLs constructed by other websites, which I have zero control over, aren't made to a certain standard. Yet the links work fine in that people can just click on the link and go to it. Who's wrong and why should I care if I can't control it and everything works fine anyway? You can control it, if you want to: it's not that hard to fix the broken URL:s on your side, using the same rules as web browsers use to parse broken URL:s. If you're not a good coder yourself, there are plenty of HTML normalization libraries that can do that for you. Whether you should care or not is another issue.

By: Mach5

Mach5 — Thu, 16 Oct 2008 08:17:48 -0800

My web 2.1 (yes, I upgraded) site, shuttercal.com, is so terribly invalidated, I'm surprised that it works. I use lots of custom attributes to make the lightbox take in all sorts of metadata. Plus, I'm lazy. If it works in FF3 and IE7, why even bother trying to get validated? I have to do some amazing hacks for transparent PNG's in IE, and I'm completely lost trying to get z-index to work in IE to make an pop up menu without more slow javascript. Front end web coding is a losing battle, I don't know how people do that as full time work. It PAINS me whenever I am forced to do it.

By: lukemeister

lukemeister — Thu, 16 Oct 2008 08:21:48 -0800

*resolves to use BDO tag in future post*

By: specialfriend

specialfriend — Thu, 16 Oct 2008 08:22:29 -0800

It's very tough to make a site that works on Internet Explorer, Firefox, and Safari. Validation is like ensuring your site works on a 4th browser that no one uses because it doesn't exist.

By: lukemeister

lukemeister — Thu, 16 Oct 2008 08:23:33 -0800

Metafilter: 29 Errors, 57 warning(s)

By: DU

DU — Thu, 16 Oct 2008 08:24:49 -0800

I don't understand all this "what does validation buy me" or "clients don't pay for standards" talk. Maybe the web design world doesn't work like the programming world, but here's why *I* use standards: So I can re-use components without worrying that it only worked in a particular context. For instance, a few years ago I made a site that I used (briefly) to document some science projects. My format was pictures on the left and explanation on the right. It was easy to make that work on an entry-by-entry basis, but harder to make it work for all pictures and explanations. But eventually I got it and I could make a template. I spend a little (of my own) time up front and I save (the client's) time on the back end.

By: Brandon Blatcher

Brandon Blatcher — Thu, 16 Oct 2008 08:26:07 -0800

here are plenty of HTML normalization libraries that can do that for you. Links please? If it's a matter of plugging in code to rewrite links automatically, that at least seems viable. Is this something that's often done with websites? Is there a server big server penalty in terms of cycles and processing?

By: adipocere

adipocere — Thu, 16 Oct 2008 08:36:31 -0800

Uh, I will weigh in on this entire topic at greater length, but I'll throw in a small bit for the derail: some URLs cannot be fixed. I have had situations in which I have properly escaped said URL only for that URL to no longer work. I first ran into this around 2003. Sometimes it is an ancient server on the other end. Sometimes it is some poorly-considered script on the other end which is thrown into confusion by a properly escaped URL. Either way, you cannot blanket fix these situations and must live with invalidating URLs if you want the actual part where someone clicks your link and gets results.

By: smackfu

smackfu — Thu, 16 Oct 2008 08:37:37 -0800

So I can re-use components without worrying that it only worked in a particular context. Validation doesn't ensure that.

By: -t

-t — Thu, 16 Oct 2008 08:39:12 -0800

I still use URL Cleaner by Dan Benjamin over at hivelogic. I'm not sure if he is still offering it or not, I couldn't find it on his site.

By: DU

DU — Thu, 16 Oct 2008 08:44:18 -0800

Validation doesn't ensure that. You are absolutely right. However: 1) I'm explaining why validation against a standard is a good idea in general. It's because having a standard means you can re-use things. 2) If my page doesn't work but it does validate, at least I know where the problem lies. Either a design error or a browser rendering error, not syntax.

By: Kutsuwamushi

Kutsuwamushi — Thu, 16 Oct 2008 08:53:07 -0800

I've never even tried to create a "professional" website; all of the pages I've written have been much less complicated than that. And yet, it still seems like there's always at least one element that isn't properly supported in some popular browser, forcing me to give up on the idea or use a clumsy hack or... give up. And that's pages that do validate. Until being standards compliant means that my pages probably won't break, there's not much incentive to make it a priority. /n00b perspective

By: Artw

Artw — Thu, 16 Oct 2008 09:00:10 -0800

Validation is way under "making it work" and "getting paid" I'm afraid.

By: Artw

Artw — Thu, 16 Oct 2008 09:02:07 -0800

That said, I'm getting increasingly weirded out by the number of SEO buggers pushing me to do things that I've been pushing for years anyway.

By: Robert Angelo

Robert Angelo — Thu, 16 Oct 2008 09:04:20 -0800

Only 4.3% validate? We are "shocked, shocked" as Captain Renault would say.

By: Artful Codger

Artful Codger — Thu, 16 Oct 2008 09:06:52 -0800

If the leading browsers can't get together on a common set of standards... OK yes web standards are good and important, and we follow them as much as possible, including accessibility, but as professionals we have a job to do. eustacescrubb > [clients] care that their customers can all see and use the site; they care about Google ranking; they care about functionality. If validation takes more time and effort and restricts what can be done design-or-function-wise but offers no benefit to the client's bottom line, then the client won't pay for it. Exactly. Even so, it sometimes seems that we spend up to a third of development time trying to accomodate browser differences, or sort out browser-specific bugs.

By: ardgedee

ardgedee — Thu, 16 Oct 2008 09:12:40 -0800

You can (and should) use conditional comments to fraction off IE-specific markup. Among other things, it allows you to write a global CSS document that every DOM-compatible browser (everything except MSIE) understands, and then write an addendum CSS document that only IE 5 (or 6 (or 7 (or combinations of them))) will see, load, and interpret. No more CSS hacks, no more invalid or bizarre markup that all browsers and validators have to endure for the sake of MSIE. There is a similar provision for Javascript. These are two of the biggest favors Microsoft ever gave web developers, and hardly anybody knows about them.

By: bonaldi

bonaldi — Thu, 16 Oct 2008 09:13:15 -0800

effbot your English doesn't validate. The checker says: URL:s wtf

By: ardgedee

ardgedee — Thu, 16 Oct 2008 09:14:31 -0800

Sorry, the Javascript link is wrong. Go to Conditional Compilation of Javascript instead.

By: blue_beetle

blue_beetle — Thu, 16 Oct 2008 09:27:11 -0800

You know who else never, ever wrote valid HTML, never put correct MIME types, never used smart-quotes properly, and NEVER WROTE bug-free javacript? That's right. Hitler. Draw your own conclusions.

By: OverlappingElvis

OverlappingElvis — Thu, 16 Oct 2008 09:27:38 -0800

Validation is way under "making it work" and "getting paid" I'm afraid. Yeah, this is unfortunately where I stand as well. Standards have done very good things for the Web as a whole, I think, if for no other reason than making browser designers start thinking about moving away from platform-specific implementations of features. I think that simply having standards out there will go a long way towards preventing another browser war a la Netscape/IE. That being said, I don't really care if my work is valid or standards-compliant per se as long as it works on FF2/FF3, IE, and Safari on both PC and Mac. I don't think I've ever put out a site that validates properly, but if it works for my audience I'm not going to lose any sleep.

By: rokusan

rokusan — Thu, 16 Oct 2008 09:40:42 -0800

Krinklyfig nails it in one. Very few people care, and of those that do, many are generally just circle-jerking their own self-controlled websites. It's very easy to make anything and everything "just so" when you only need to satisfy an audience, QA department, account rep, client, and review board that's made up of you, and only you. And, of course, this survey also doesn't factor in the ever-present "I am aware that this sub-part does not validate by that reference, and here is why we made that decision, and here is how it performs in the browsers this project cared about." criteria. Validation and standards are not inherently bad. They're just overhyped and too often oversimplified down to a true/false measure... like this survey.

By: burnmp3s

burnmp3s — Thu, 16 Oct 2008 09:47:28 -0800

Maybe the web design world doesn't work like the programming world, but here's why *I* use standards Standards work better in the programming world because they are almost always enforced better than they are in HTML. When I write code that has a minor syntax error, my compiler's parser throws a handy set of errors and/or warnings that tells me specifically what I'm doing wrong. A web browser, on the other hand, is specifically designed to always try to display a page, no matter how horribly mangled the content is. Even in the programming world, if standards aren't enforced in real life they tend to not be followed. SQL, for example, has a nice standard that database vendors ignore large parts of and add extra features on top of, which results in a lot of SQL code that doesn't follow the official standard. I think HTML pages would be a lot more standards compliant if the standards and browsers had been created today instead of more than a decade ago. For a long time the mantra around implementing protocols and standards was Postel's Law: be conservative in what you do, be liberal in what you accept from others. The problem is that using a more "robust" parser will often result in very wrong output with little or no warning, and will not do much to deter people from giving it bad data, and people have recognized the impacts of these problems in established standards like HTML. The more recent XML standard is a case where the overall programming community has explicitly called for less robust parsers by writing parsers that only accept valid XML, in order to promote proper implementations. Obviously being able to handle errors gracefully is still important, but there is a lot less tolerance in the community as a whole for non-compliance to standards.

By: cillit bang

cillit bang — Thu, 16 Oct 2008 09:55:05 -0800

then validation fails because the URLs constructed by other websites, which I have zero control over, aren't made to a certain standard. $link=str_replace('&','&',$link); or $link=htmlspecialchars($link); Or the equivalent in your language. Ampersands in links (and any other attributes and most pieces of text) should always be escaped in this way, and this will not break the links in any browser I know of.

By: kirkaracha

kirkaracha — Thu, 16 Oct 2008 10:02:13 -0800

My sites generally validate almost automatically because I write standards-based code as part of the production process. It's not an extra step, and it doesn't take extra time. I like having valid code because I feel that it helps reduce cross-browser differences. I don't usually validate every page as part of launching the site, but when I spot-check the sites usually validate.

By: cillit bang

cillit bang — Thu, 16 Oct 2008 10:02:24 -0800

Hang on, since politicalfilter accepts HTML from users rather than just links, you need something like HTML Purifier, which will process the HTML and fix the dodgy links, as well as filtering various scripting attacks and doing other code tidying.

By: smackfu

smackfu — Thu, 16 Oct 2008 10:04:27 -0800

Also nowadays with everyone using javascript libraries, all your dependencies have to be valid too. And all the ads you're including better be valid too.

By: Artw

Artw — Thu, 16 Oct 2008 10:08:54 -0800

I'm unconvinced that messing about with ampersands in links benefit anyone whatsoever.

By: Artw

Artw — Thu, 16 Oct 2008 10:28:51 -0800

(It's exactly that kind of thing that makes working in XML or XSLT a pain in the ass)

By: Dreama

Dreama — Thu, 16 Oct 2008 10:30:07 -0800

My blog, which has a valid template, fails in part due to a Flickr generated badge in my sidebar and an embedded YouTube video. Standards compliance is going to come from the top down. YouTube belongs to Google. Flickr belongs to Yahoo. If the content they're putting out isn't valid, why are all the hobbyists and smaller companies (like Last.fm and AllConsuming, whose badges in my sidebar are also not valid for my transitional doctype) supposed to bother with compliance, again?

By: sonic meat machine

sonic meat machine — Thu, 16 Oct 2008 10:52:41 -0800

My personal sites validate. My job-related sites, however, don't—because my bosses want it to look just so to match the piss-poor graphic design job they paid some hack to work up for them.

By: cillit bang

cillit bang — Thu, 16 Oct 2008 11:00:45 -0800

I'm unconvinced that messing about with ampersands in links benefit anyone whatsoever. It's about ambiguity. The code: <a href="http://example.com/query?q=2"234"> Should always mean the URL http://example.com/query?q=2"234, but since most HTML coders don't bother to escape their ampersands, the browser has to guess whether you might have meant http://example.com/query?q=2"234 instead*. And the logic to do that guesswork has to be incorporated into every single HTML interpreter. Alternatively, everyone could follow the bloody standard in the first place and we wouldn't get into this mess. The horse has long since bolted for HTML, but XML has very strict rules that forbid any parser from doing any kind of guesswork, for exactly this reason. (* this example might sound contrived or rare, but the way RSS fails to address it is a big part of the reason Atom exists and has been widely adopted)

By: caution live frogs

caution live frogs — Thu, 16 Oct 2008 11:03:57 -0800

specialfriend wrote "It's very tough to make a site that works on Internet Explorer, Firefox, and Safari." No, it isn't. It's extremely easy to make a site that works on IE (any flavor), Firefox, Netscape, Lynx, Amaya, Webkit, what have you. It is very tough to make a site that works well across browsers if you wish to have any semblance of control over the visual display of the content, or want to add basically anything except images. Plain old HTML is dead simple. The problem starts when you try to make a consistent look and feel across browsers. A major part of this problem is the plain fact that up until fairly recently, some of the major browser vendors (*ahem Microsoft cough cough*) didn't give two shits about standards, and expected web authors to find hacky fixes. Things are beginning to get better. You have three choices, as I see it. 1) Work towards validation only: Accept that you really have no control over the display of content, and use nothing more than plain unstyled text and perhaps static images, letting the browser display them as it will. 2) Aim for consistency at the expense of validation: Work damn hard trying to make things look the same in as many places as possible, using whatever measures necessary. 3) Do your best to generate valid code, that shows as much cross-browser consistency as possible, starting with the code itself and building upwards (rather than thinking about graceful degradation, go the opposite direction: Content first, presentation as supported) while at the same time pushing as hard as possible to get browser vendors to agree on at least basic standards. 1 is giving up and going backwards. 2 is a pain in the ass, now and forever, with a large possibility that each new iteration of a browser will require a huge time investment to test and patch against. 3 is perhaps harder now, but ought to be the easiest long-term solution. But what do I know. I'm just a guy who writes web pages because I can, and reads A List Apart on occasion to see just how much of an amateur I really am.

By: Nelson

Nelson — Thu, 16 Oct 2008 11:07:51 -0800

In related news, 99.7% of the web works pretty well in all major browsers.

By: Artw

Artw — Thu, 16 Oct 2008 11:16:24 -0800

Yeah, that's exactly the kind of utterly contrived problem you get by treating your HTML as XML. Putting an encoded character in there is ugly and a pain in the ass, and then you get into problems when your encoding and unencoding things and end up with ridiculous shit like & turning up in code. I guess really the URL should be CDATA, but obviously that's not going to fly.

By: Artw

Artw — Thu, 16 Oct 2008 11:17:16 -0800

A major part of this problem is the plain fact that up until fairly recently, some of the major browser vendors (*ahem Microsoft cough cough*) didn't give two shits about standards, and expected web authors to find hacky fixes. Netscape 4, how quickly you are forgotten.

By: Skorgu

Skorgu — Thu, 16 Oct 2008 11:20:11 -0800

Frameworks should handle validation as shipped, by default, including user-provided data, links, etc. That there are some that don't do this is why we have XSRF and XSS attacks.

By: sciurus

sciurus — Thu, 16 Oct 2008 11:29:13 -0800

Reset.

By: Civil_Disobedient

Civil_Disobedient — Thu, 16 Oct 2008 11:34:32 -0800

XHTML deal-breakers:

All content has to be dished out as application/xhtml+xml -- or application/xml for older versions of IE
I have to encode all my ampersands. EVERY SINGLE SOLITARY AMPERSAND. Like this & this & this? That's 8 extra bytes. And you better do it, otherwise your browser will choke and your house will be catch fire and your family will die.
< > & " and ' are the only guarnateed entities. Bye-bye   &hearts; × and a whoooole lot of others.
Any javascript on the page? Better make sure to wrap it in a <[CDATA[ block or suffer the browser's wrath when it hits its first AND condition.
Oh yeah, and document.write doesn't work. No Google Adsense for you (without tricks).

HTML (strict) deal-breakers:

<input>, <br>, and a bunch of other tags aren't closed! So any <br /> or <input ... /> will be invalid. Hope you're not using any server-generated code (no Java framework currently gets this right, for example, and I'd bet hard money no .NET platform can do it right, either).
Bye-bye <blockquote>, <i>, <b>, <u>, <s>.

By: cillit bang

cillit bang — Thu, 16 Oct 2008 11:39:04 -0800

Yeah, that's exactly the kind of utterly contrived problem you get by treating your HTML as XML. The problem is that treating HTML as "HTML" requires putting complex algorithms into the intepreter that try to guess at exactly which combinations of ampersands, characters and semi-colons should be counted as entities and which should be left as is. That's your encoding problem right there. Whereas if everyone was standards compliant there'd be exactly one rule for everyone to follow. (I don't actually support the arbitrary standards compliance for HTML that Opera et al are advocating, since the problems caused by lack of validation or lack of well-formedness are a tiny, unimportant subset)

By: Artw

Artw — Thu, 16 Oct 2008 11:45:26 -0800

Visual Studio will autocorect your <br> tags to </br>. And yes, all of this is very annoying when trying to work to standards, and you get the general impression that the people creating the standards are far more into some weird fetishism of the form than actually doing anything that's useful, which contributes in part to my fuck you attitude to validation.

By: lodurr

lodurr — Thu, 16 Oct 2008 11:48:58 -0800

Geez, C_D, that's grim. (I used to know all that shit when I first started out in the business, back when dinosaurs ruled the earth. But I gave up trying to keep up years ago and now I just make stuff that works. I'm a really bad person, I guess....)

By: Artw

Artw — Thu, 16 Oct 2008 11:58:28 -0800

Also what fucking idiot decided that the widths of elements would not include padding and borders? That opens up a whole heap of problems, and completely fucks up anything percentile without a lot of pissing around. I'd like to find whoever made that decision and beat them over the head with a book of CSS hacks.

By: mandal

mandal — Thu, 16 Oct 2008 12:12:11 -0800

Not obligatory.

By: Artw

Artw — Thu, 16 Oct 2008 12:23:17 -0800

You might want to talk to every client I've ever had about that.

By: sexyrobot

sexyrobot — Thu, 16 Oct 2008 12:37:26 -0800

wow. wow. seriously? seriously??? you'd think that a whole gang of web developers would get it, but apparently not...(seriously? really?) the internet isn't about conforming to any standards...the internet is about AND AND AND AND AND. AND HTML4.0 works, AND CSS works, AND tag clouds work, AND embedded video works, AND, hell, even HTML 1.0 still works and that expired YEEEARs ago. Validated? validated by who? who's that? never heard of him, why should i care? if only 4.3% of sites pass validation, then it sure don't look like validation is very valid. i'd suggest that it should be ejected from the web, but that's not really necessary, even if it's useless.

By: Eideteker

Eideteker — Thu, 16 Oct 2008 12:52:57 -0800

My site failed because when you compose a basic HTML page in notepad, it doesn't specify a UTF-8 encoding. Sorry, man. When I learned HTML back in 1993, we knew nothing of this "UTF-8". It was just HEAD, TITLE, BODY, done.

By: Bora Horza Gobuchul

Bora Horza Gobuchul — Thu, 16 Oct 2008 13:18:37 -0800

We've talked about this before. I'm surprised that the percentage of valid websites is as high as 5% - as mentioned, it could be even higher, were it not for generated content breaking a number of pages. But really, where's the shock? Most people are barely competent, in almost every field. (There are exceptions in the higher-level professions, where results really matter, but on the whole the larger the field the more niches exist for people who barely know what they are doing.) This is especially true in web development, where everyone considers themselves a creator, or at least a contributor. It's made somewhat better, at least in theory, by content management tools and design, if most of those tools weren't half-assed to begin with (Myspace and DreamWeaver, I'm looking in your direction). Very few people taken from the street would consider themselves prepared to write, illustrate and bind a book. But everyone believes they can make a web page. Don't get me wrong: I'm all for open access. But standards have a purpose. They might be arcane, they might be fussy, they might be a pain to implement, but they serve a very logical and ultimately very useful purpose. And, turning to my peers for a moment, the industry has not been served well by the teaching profession. I teach fulltime and continuing education courses at a polytechnic, and have to fight instructors in other departments who still teach their students to use the font tag, with tables for presentation. In development, the credo has been the trend has been "make it work in IE, whatever it takes". Which is like bashing a square peg to fit into a round hole when the hole itself is wrong. Doing it right the first time is not difficult. And, slowly but surely, developers are getting the idea: I was stunned to learn that the MSN.com index page validated as strict. Once you do start to build valid code, everything else - CSS, DOM JavaScript - makes a lot more sense, and actually works. Yes, IE6 continues to be a massive spear in the side of suffering web developers. Conditional comments work wonders; longer term, the slow, gathering, steady pressure to drop IE6 support next year may have some effect.

By: Artful Codger

Artful Codger — Thu, 16 Oct 2008 13:47:04 -0800

conditional comments in IE are an example of the underlying browser problem- solving a kludge with another kludge. BTW there are other (harder) ways to put in conditional actions depending on browser. They also happen to be standards-compliant... just sayin'. Also possible to handle server-side. > These are two of the biggest favors Microsoft ever gave web developers, and hardly anybody knows about them Favours? (head asplodes) If you're really into self-abuse, try outputting HTML from MS Word... gaaaahhhh.

By: tkchrist

tkchrist — Thu, 16 Oct 2008 14:03:49 -0800

Funny thing is nobody gives a shit about any of this anymore.

By: joeclark

joeclark — Thu, 16 Oct 2008 14:29:37 -0800

Rational and competent developers write valid HTML and CSS first, then tweak as necessary. In unusual cases the tweaks unavoidably include invalid elements, attributes (hasLayout, tabindex=-1), and declarations. It's oft-repeated in fora like this one that I Just Have to Get This Fucker Working, but the way you do that is to start with correct code. You should never be surprised that this fucker isn't working when it uses invalid code. And apart from maybe acronym/abbr in IE6 and longdesc, I fail to see how using valid "tags" will cause things not to work in modern browsers. Nonetheless, we are at the point where unescaped ampersands and the EMBED tag (sic) mean nothing.

By: TheOnlyCoolTim

TheOnlyCoolTim — Thu, 16 Oct 2008 14:52:26 -0800

This validator is invalid: Error Line 10, Column 6: element "BLINK" undefined. Quod erat demonstrandum.

By: turgid dahlia

turgid dahlia — Thu, 16 Oct 2008 14:53:37 -0800

Oh no Metafilter! "64 Errors, 123 warning(s)"

By: Civil_Disobedient

Civil_Disobedient — Thu, 16 Oct 2008 15:53:44 -0800

now I just make stuff that works. I'm a really bad person, I guess... Nothing wrong with the pragmatic approach. They might be arcane, they might be fussy, they might be a pain to implement, but they serve a very logical and ultimately very useful purpose. Really? Then can you please explain to me what the logical and ultimately very useful reason why this is invalid?

<form>
  <input type="text" />
</form>

A <form> and an <input> field... what could be simpler? Simple and invalid. And how about the target attribute? The W3C in their infinite wisdom decided that the handy-dandy target attribute of a link is deprecated. So how do you open links up in new windows/tabs? Answer: Javascript. Any way to do it without javascript? Answer: nope. Next question?

By: srboisvert

srboisvert — Thu, 16 Oct 2008 15:56:58 -0800

Screw validation. My website visitors can pay for their own damn parking.

By: maxwelton

maxwelton — Thu, 16 Oct 2008 16:14:26 -0800

Validation would be fine if the actual standards weren't written by a bunch of hose-heads who have never, ever been paid to design and construct a commercial website. Yeah, "fluid design" websites are great, and I try to do them when I can, just as I try to make my sites validate. Alas, most clients, do, in fact, want "pixel perfect" designs. Yeah, that's just evil and blah blah blah. Too fucking bad, that isn't going to change. Jakob whatever his name is can bite my unvalidated sack.

By: Artw

Artw — Thu, 16 Oct 2008 16:34:03 -0800

So how do you open links up in new windows/tabs? Answer: Javascript. Any way to do it without javascript? Answer: nope. Next question? JavaScript is teh evils you terrible accesibility breaker you. Clearly, like aligning things vertically, its a thing you don't need to do and shouldn't do.

By: RobotVoodooPower

RobotVoodooPower — Thu, 16 Oct 2008 17:41:43 -0800

And 99.9999% lack properly formed OWL semantic web tags! Impudent charlatans, I say, all of them!

By: signal

signal — Thu, 16 Oct 2008 18:14:45 -0800

I only care about validation when something stops working and I can't figure out why. Then, I validate, and about 50% of the time fixing the validation error fixes the bug.

By: furtive

furtive — Thu, 16 Oct 2008 18:34:48 -0800

When I write html from scratch it's always 100% valid. When I inherit sites, if the code is invalid I don't lose any sleep over it as long as it renders in IE/FFox.

By: Bora Horza Gobuchul

Bora Horza Gobuchul — Thu, 16 Oct 2008 19:33:37 -0800

Really? Then can you please explain to me what the logical and ultimately very useful reason why this is invalid?

<form>
  <input type="text" />
</form>

Sure. Just off the top of my head:

Your input text box doesn't have a label. The person filling out the form has literally no idea what you're asking for.
You're not using an action attribute for the form, and don't have a submit button, so you're not using PHP, Perl, or some other server-side language. Instead, the assumption is that you're validating the form via JavaScript. Fair enough. But there's no id on the input box itself for the JavaScript to hook into.
An id would also work with the missing label to help make the form accessible, which is important - kind of like making a wheelchair ramp to the front door of a business or public institution.

Hope this helps!

By: [insert clever name here]

[insert clever name here] — Thu, 16 Oct 2008 19:47:45 -0800

I work in a shop where no one validates their code (except me). What that means is that every person writing a website has their own crazy style of code that is nearly impossible to interpret when another designer picks it up. '20 nested divs inside an a tag? Great idea!'. Validation isn't perfect but it does help weed out really bad code that is there for the sake of being bad (or a lack of understanding). Now, that being said, when I'm done, my pages don't always validate. But that's because I've made a conscious decision to deviate from the standards for some reason or another. I just used lightbox++ on a site, and had to give up a perfectly valid page for the attributes to make flash work. But that's okay, the end result was more important. The nice thing about page validation too is that it ensures longevity of a page. Not just in browser rendering but in search engine placement in regards to using the right code in the right places. Of course there are exceptions where things that were valid end up being depreciated, but on the whole, you're safer future-proofing a site by making sure its using valid markup. Also, if clients won't pay for the time to do it right, that falls squarely on the shoulders of said developer for not making the business case for doing it right. Or, you know, of course why should you have to explain it if you're doing it right the first time?

By: signal

signal — Thu, 16 Oct 2008 19:49:29 -0800

Yeah, "fluid design" websites are great, No, they suck, as the line length is either too big or too small, depending on the size of the browser window. The web is about typography, not code.

By: Bora Horza Gobuchul

Bora Horza Gobuchul — Thu, 16 Oct 2008 20:11:21 -0800

The web is most definitely not about typography. If it is anything (and I don't find this definition particularly appealing either) the web is about the semantic markup of data. However, I would give you the fact that support for typography on the web has been middling to poor until recently. Fortunately, the emergence and slow adoption of CSS 3 standards, particularly in regards to web fonts, is very promising (albeit long, long overdue).

By: cillit bang

cillit bang — Fri, 17 Oct 2008 01:59:30 -0800

Really? Then can you please explain to me what the logical and ultimately very useful reason why this is invalid? Because it doesn't have an action attribute. Sure, most browsers will default to behaving the same way as if you'd inserted an empty action attribute, but since that behaviour isn't mentioned in the spec, you have no business relying on this happening, unless you're happy that your site will only work in whatever subset of browsers you know to behave this way. Which I guess you are.

By: signal

signal — Fri, 17 Oct 2008 07:38:33 -0800

Bora Horza Gobuchul: Typography != choosing fonts. Rather it's presenting textual and graphic information in a legible, useful, meaningful and attractive way. This is the main function/problem of the web, and fluid web sites mostly suck at it, present company excluded, of course. I think the limited font choices on the web are a feature, making it much more readable. The A List Apart article you link to shows how unlimited font choices can lead to some fairly awful designs in the hands of the typographically uninformed, more of a threat than an opportunity. Better 'support' for typography would include having decent justifying and hyphenation in browsers, being able to use true caps, consistent indentation, etc.

By: Artw

Artw — Fri, 17 Oct 2008 08:25:45 -0800

The web is about content, and web standards should be about delivering that content in the best way. When they become about pedantic code smuggery then they fail utterly.

By: joeclark

joeclark — Fri, 17 Oct 2008 10:18:48 -0800

Signal, I assume you meant "being able to use true small caps." I don't see how indention is not "consistent" on the Web. You can control the problem of marathon line lengths with max-width (requiring a JS fix for IE6). Artw, for "pedantic code smuggery" I will read "correct semantics."

By: Artw

Artw — Fri, 17 Oct 2008 10:38:10 -0800

Hey, I'm all for nicely structured code and accesabilty and all that, but encoding mid-URL ampersands? That's bullshit. Making a flap over people using I instead of EM is another one that's always seemed damn silly, no one in the real world cares if you made something italic using I or EM, and TBH I couldn't really tell you the difference between something that has been semantically emphasized and something that is semantically strong.

By: Civil_Disobedient

Civil_Disobedient — Tue, 21 Oct 2008 04:00:13 -0800

Hope this helps! No, you moron, you completely missed the point. Because it doesn't have an action attribute. Argh, idiots everywhere! The big, large, flaming tree of a point that the awfully fucking literal folks around here seem to have missed is that the reason the code above is invalid is because you cannot enclose an input tag (an inline element) inside of a form tag. The code above should fail in HTML 4.01 (Strict). You must use a block level element for the form to be valid. (What's that...? An ACTION attribute, you say? NO KIDDING!?) "No problem! Just add a <fieldset> around it," you exclaim! What? I'm sorry I can't hear you because I'm laughing so loud. The fieldset? One of the most inconsistently half-assed HTML tags!? Oooh yeah, let's add a <legend> tag to the mix as well, then slit our wrists playing with positions and borders for the next two weeks. Can we please? Clearly, like aligning things vertically ARGH! Exactly! To the W3C, even the concept of height scares them. It's a page of infinite goddamned length! "Uh, well, no your Majesty, actually it's a PDA with a resolution of 600x400... or it's a computer LCD with a resolution of 1280x1024... or it's a piece of paper with a resolution of 8.5" x 11"..." The W3C has lost all notion of sanity at this point. Just code what works.