Recomendations for doctype

jfleck25

What would you recommend for a doctype for my website? http://www.sourcevibe.com

traq

Is there something wrong with the one it has?

Weedpacket

I can't recommend any doctype, since right now your document doesn't qualify for any of them. E.g. http://validator.w3.org/check?verbose=1&uri=http%3A%2F%2Fwww.sourcevibe.com%2F

dalecosp

If you want to fix the errors that Weedpacket refers to, keep the one you have; otherwise I'd switch to the one that gave the least number of errors... 😉

bradgrafelman

HTML entities in all caps, tables used not for tabular data but for design, inline styles strewn all over the place... I think the 1990s are calling and asking for their webmaster back!

dalecosp

bradgrafelman;11032977 wrote:
inline styles strewn all over the place...

sigh ... I've a bit to do on that one, myself....

Rivet

dalecosp;11032963 wrote:
If you want to fix the errors that Weedpacket refers to, keep the one you have; otherwise I'd switch to the one that gave the least number of errors... 😉

I'd agree with that. I've come to like HTML 5's doctype - it's simple, works across the board it seems, and since I've started digging into HTML5 it's not that complex at all except for a few deprecations which are easy to work around. I've really come to appreciate W3's code analysis, too; using it more and more now.

Twayne`

traq

At the end of the day, a valid doctype triggers "Standards Mode" in browsers, and (almost always*) nothing else.

From a practical standpoint, and in most situations, it is therefore not very important which one you choose. (That means that HTML5's doctype - being the shortest and therefore easiest to type - wins out).

* There are situations where it does matter, but I will be surprised if anyone knows of anything more than "edge" cases.

johanafm

traq;11033021 wrote:
At the end of the day, a valid doctype triggers "Standards Mode" in browsers

Which today means HTML 5, since HTML 5 supersedes previous html versions. Wether you provide a valid html doctype for 4.01 (including the dtd), or simply <!DOCTYPE html> makes no difference - it will still be considered an HTML 5 document - Assuming the browser supports HTML 5. Since HTML 5 is based on parsing rules instead of SGML dtds, it will always be processed in the same way and the dtd will be ignored.

As far as the doctype declaration goes, there would be two exceptions to the above. The first is if the browser doesn't recognize <!DOCTYPE html> as a valid doctype declaration. Which in turn means the browser do not recognize HTML 5 even partially. And the second would be to provide an incorrect doctype declaration so that the document cannot be considered html.

The choice then boils down to serving your html 5 document as an html or an xml serialization.

If you serve it as xml, you do not need the doctype declaration at all since it will be ignored. But you may leave it there. The only thing used to decide that this is an xhtml 5 document is the mime type: application/xhtml+xml. You will need an xml declaration, which must be the very first thing in the document (BOM ok?): <?xml ...>. You will also need to provide a default namespace on your root element: <html xmlns="go look it up somewhere">. Thus it would look something like

<?xml ...>
<!DOCTYPE html>
<html xmlns="go look it up">

and be served as application/xhtml+xml. The doctype declaration may be left out, but if it is present, DOCTYPE must be in upper case letters.

While the html serialization would be

<!DOCTYPE html>
<html>

served as text/html. And the doctype is still there only for legacy reasons so that some browser doesn't jump into IE5 mode, which it may do if your doctype declaration is incorrect. But you would not intentionally do that, would you?

As for some of the differences that I find may be of importance:
- No use of document.write in xml documents
- No use of noscript elements in xml documents
- cAsInG matters in xml documents.
- no namespaces in html

Rivet

johanafm;11033023 wrote:
Which today means HTML 5, since HTML 5 supersedes previous html versions. Wether you provide a valid html doctype for 4.01 (including the dtd), or simply <!DOCTYPE html> makes no difference - it will still be considered an HTML 5 document - Assuming the browser supports HTML 5. Since HTML 5 is based on parsing rules instead of SGML dtds, it will always be processed in the same way and the dtd will be ignored.

As far as the doctype declaration goes, there would be two exceptions to the above. The first is if the browser doesn't recognize <!DOCTYPE html> as a valid doctype declaration. Which in turn means the browser do not recognize HTML 5 even partially. And the second would be to provide an incorrect doctype declaration so that the document cannot be considered html.

The choice then boils down to serving your html 5 document as an html or an xml serialization.

If you serve it as xml, you do not need the doctype declaration at all since it will be ignored. But you may leave it there. The only thing used to decide that this is an xhtml 5 document is the mime type: application/xhtml+xml. You will need an xml declaration, which must be the very first thing in the document (BOM ok?): <?xml ...>. You will also need to provide a default namespace on your root element: <html xmlns="go look it up somewhere">. Thus it would look something like
<?xml ...>
<!DOCTYPE html>
<html xmlns="go look it up">
and be served as application/xhtml+xml. The doctype declaration may be left out, but if it is present, DOCTYPE must be in upper case letters.

While the html serialization would be
<!DOCTYPE html>
<html>
served as text/html. And the doctype is still there only for legacy reasons so that some browser doesn't jump into IE5 mode, which it may do if your doctype declaration is incorrect. But you would not intentionally do that, would you?

As for some of the differences that I find may be of importance:
- No use of document.write in xml documents
- No use of noscript elements in xml documents
- cAsInG matters in xml documents.
- no namespaces in html

Good post; thanks for the input.

Rivet`

dalecosp

traq;11033021 wrote:
* There are situations where it does matter, but I will be surprised if anyone knows of anything more than "edge" cases.

That might depend on whether you consider hundreds of thousands of SOHO (or cubicle-based?) drones who are still on WinXP with IE7/8 as an edge case, wouldn't it?

traq

dalecosp;11033053 wrote:
That might depend on whether you consider hundreds of thousands of SOHO (or cubicle-based?) drones who are still on WinXP with IE7/8 as an edge case, wouldn't it?

I haven't tested IE6, but everything I've read says the HTML5 doctype does trigger standards mode. I have tested in IE7, and it triggers standards mode as expected.

If you're talking about the HTML5 elements (e.g., <header>, <nav>, <main>, etc.) being broken in IE<9, yes, I'm aware of that and I don't consider it an "edge case." But they don't work regardless of standards/quirks mode (not without help), and the doctype doesn't affect that anyway.

dalecosp

Good point; my main issue seems to be that each version's interpretation of some details differs unless you force them to use IE=EDGE, if I recall correctly.

traq

yeah.. it's a big mess. But what are ya gonna do.