Somebody just showed me the sIFR method for using flash to create complex fonts while keeping search engine and non-flash supporting browser support. Read it for yourself on mikeindustries.com.

There are a lot of claims made for this method, although amusingly the guy doesn't set the content type on his xhtml, so it's a fair bet he doesn't know much about XHTML, this tag was missing.

<meta http-equiv="Content-Type" content="text/html; charset=iso-8859-1" />

Despite his claims, this stuff is just marginally xhtml compliant, if these pages were served with correct mimetype of xhtml+xml they would not work, but since they are just regular text/html they do work, so it's more accurate to call this html compliant than xhtml compliant. The idea of using javascript to get xhtml 1 non xhtml 1 mimetype validation is one I use too, but it's a total hack, not a real solution to the problem.

It's a fairly slick idea, and well executed, though I'm surprised they made a very simple error, Opera doesn't support onload events for security concerns unless they are called in the body tag I think, so of course this script doesn't, or shouldn't, work in opera. It probably doesn't work in safari though I can't say for sure, worth checking out though.

What's odd about it is that you can do exactly the same thing by just having defaults in first your object tag, then your embed tag, then you have the html you want the search engine or viewer to see, none of this is actually necessary.

That looks like this:
<object>.....
<embed>....<h1>the raw html</h1></embed></object>

that default has always been built in, I'm not positive if it cascades
correctly with 2 inner defaults but it would be easy to find out.

This doesn't validate, but if one is honest, using javascript to write out invalid code is not an actual solution either, if the javascript written code were added to the validator the page would fail too, this is why real xhtml doesn't support any type of document.write or innerHTML, which is by the way not even a real standards supported method, it's just something browser makers keep supporting because there's no real substitute, but it's not in fact DOM compliant if I remember right, so this is not an actual standards compiant method, though it does work well.

The swf used is very big, 23 kB, that's about 5-7 seconds over dialup modem to achieve what is basically just a simple cosmetic affect, the js is 10 kB, but could have about 3 kB stripped off of comments, that's 30 kB total to modify about literally 1 kB of data, which is what it would take to create a full CSS version of two css text items. But you'd still need more css ontop of that to handle the defaults, as I often find to be the case, when something sounds too good to be true, it usually is. 30kB is a lot of data to just print out a tiny amount of text on a page. If I'm going to use flash, I usually try to get as much display value as possible for the file size.

But it's another cool affect, I see these a lot, alistapart is full of them, slick stuff that always has one type of problem or other when you look at them more closely. Despite these practical problems, it's always nice to see what people are playing with, personally I find using flash is just pain, it's harder to edit, it's harder to maintain, and it's a lot bigger, and requires hacks to get search engines in, I gave up on flash a long time ago although I see there are times when it's a good addition to a page.
Back to top

<meta http-equiv="Content-Type" content="text/html; charset=iso-8859-1" />
would be wrong to use if you're serving xhtml you wouldn't want to really serve it as text/html. i believe it should fall back to utf-8 by default.

also he's not very huge on standards..see
here

regardless there are many others that use this technique and their pages are compliant. see: www.designbyfire.com
Back to top

btw read all the comments, they sort out a bunch of the problems...the project is still in BETA. the mac/ie5 issue is addressed
Back to top

Yes, I've read that before, it's just the same old lame argument coming from someone too lazy to do his job right, and wanting to make excuses about that, I've seen that before, basically every incompetent web designer I've ever dealt with gives some form of that argument.

the bottom line is that creating a valid html 4.01 transitional document is about as easy to do as anything you can do, and it's easy to debug.

The whole xhtml thing is besided the point, I've seen and read excellent arguments that, in my opinion, correctly point out that xhtml, unless served as mimetype application/xhtml+xml is actually invalid html, the more I read about that the more I agree with this point.

I'm definitely not a validation type freak or anything, but I've come to realize that when you work with html and css, you simply have no other debugging tools available to you, unlike all programming languages, even javascript. It's because almost all web designers come from a design, not a programming, background, that they don't understand why your code should be, and remain, error free. It's a long term investment in the site, if you don't do it error free all the time, errors start creeping in, and they tend to be random, and cumulative, and after a while you need to rewrite the site.

As I've said elsewhere, the only thing that ESPN site is good for is as an example of how not to do web design, and how to create an inaccessible website that is going to look very cheesy very soon.

Until browsers correctly support XHTML mimetypes, currently only Opera and Mozilla/firefox do, I'm creating all my commercial stuff as HTML 4.01.
Back to top

well if you code your site as valid xhtml and go but the utf-8 charset, all you have to do is change the text/html when the day comes that browsers except the mime type.

the one thing that bothers me about xhtml is that if you're using links to any long sort of dynamic url there's a good chance it won't validate..I really still don't know how one should get around that.


I don't know how it got to this, but the sIFR validates so i don't really see the problem, or how we got this off topic. his sample page is not meant to be a valid masterpiece and the code is full of commenting and bloat, but it's that way for a reason. because it is their to educate people on how it works.

I suppose espn.com is a poor example..here's an article from the guys who launched the new compliant sprint website. they use sIFR and they explain why
www.zelph.com/archives/2004/06/20/using-ifr-at-sprint/

<added by moderator, from other thread>
oh and as i mentioned before there are numerous reasons i try to code most of my stuff in xhtml even though I don't serve it as xhtml/xml:

1. it keeps me in the habit of coding cleaner and more efficiently. which in turn allows me to more easily debug and edit code.

2. the fact that it is served as text/html is really insignificant. you just don't get the added efficiency you would have you served it as xhtml/xml. sadly though it's necessary to serve it as text/html because of the amount of users with browsers that still don't support it.

3. When the time comes that the overwhelming majority does support the xhtml/xml mimetype it will be extremely easy to change the one line of code.

4. coding in html allows others who may be working on your project to use shitty code and pretty much get away with it. xhtml coding promotes better programming IMO. the more it's used the more others will be forced to learn how to code cleanly and semantically as they are supposed to.

see: www.kurafire.net/articles/case-for-xhtml

also:
annevankesteren.nl/archives/2004/02/xhtml-versus-html

all of whom make similar points but are more in depth in their explanations.

btw i was just reading through the news today and i noticed that abcnews is now using sIFR on their website as well
Back to top

Oh, I was just responding to that link you posted, I'd read that thing too, it really struck me as kind of lame, as you say, the method you're talking about validates, so why be silly and deliberately make things not validate, seems kind of childish.

For query strings, you do this:

/yourfolder/yourpage.html?item1=dogfood&amp;item2=catfood

and so on, it's just the ampersand. This is no big deal at all, since pretty much all scripted sites are done dynamically, it makes no difference if you script in the command to write out & or &amp;

There really isn't very much difference which sign you put in, & or &amp when you're writing out the script, so doing it wrong is no harder than doing it right, so why not do it right?

I run all my client documents through a macro which strips out all non-standard html characters and replaces them with the correct character entities, just takes a button click and it's done.

People think that things like word magic quotes somehow will actually display correctly cross browser and cross os, but that isn't the case.

Anyway, let us know when you get a working sample of that method, I'd like to see a small file size version that actually works, as noted, the one above was really big, far far too big for the few strings of text it was creating. 5 seconds over dialup, not good. But I'm sure it can be done better, and much smaller, I've seen some nice flash things that are tiny, the dancing girl on the index page of this site is I think 7 kB.
Back to top

This thread was split to the programming forum, where a few more comments on XHTML occur, but mostly it moved into a discussion of using JSP or PHP.
Back to top