W3C validator interface deprecated Options

[Frederiek]

Lately, when I validate a web page at http://validator.w3.org I get the following warning :

"This interface to HTML5 document checking is deprecated.
Use the Nu Html Checker at https://validator.w3.org/nu/ directly instead."

So, time to update your bookmarks and bookmarklets.

[Christian J]

I only see that message when validating as HTML5. :-/

[pandy]

That's because it only concerns HTML5.

[Christian J]

This "Nu HTML Checker" seems confusing and unreliable. First I got this warning/error:

Warning: Using windows-1252 instead of the declared encoding iso-8859-1.
Warning: Legacy encoding windows-1252 used. Documents should use UTF-8.
Error: Bad value iso-8859-1 for attribute charset on element meta: iso-8859-1 is not a preferred encoding name. The preferred label for this encoding is windows-1252.

for using this in my test page:

CODE

<meta charset="iso-8859-1">

--even though iso-8859-1 seems to permitted by the HTML5 spec (which says "Authors should use UTF-8", not that they must): http://www.w3.org/TR/2014/REC-html5-201410...ta.html#charset (if I use windows-1252 instead of iso-8859-1, I just get a warning).

A quick web search turned up this discussion:

http://lists.w3.org/Archives/Public/www-va...13Mar/0052.html
http://lists.w3.org/Archives/Public/www-va...13Mar/0053.html
http://lists.w3.org/Archives/Public/www-va...13Mar/0054.html
http://lists.w3.org/Archives/Public/www-va...13Mar/0057.html

To quote Yucca Korpela from the last link:

If the W3C Markup Validator uses nightly
builds of editor's drafts rather than W3C Candidate Recommendation as
"the current spec", it is becoming more and more questionable to use it.

Declaring literally billions of pages invalid just because they declare
ISO-8859-1 helps nobody and makes people ask what the validator is for.
People use validators to find real problems they might have, not to get
confused about error messages that reflect just someone's opinions du jour.

I also got an error for using the MINLENGTH attribute with a INPUT type=text field, even though it's allowed in the spec: http://www.w3.org/TR/2014/REC-html5-201410...e-(type=search)

[pandy]

Told you. There won't be any true validators anymore since there won't be any DTDs. This verbose blah blah is what we'll have to put up with . Errors, mistakes or too high ambitions aside. One of the reasons i'm not p*ssing my pants over HTML5.

[Christian J]

Told you. There won't be any true validators anymore since there won't be any DTDs.

So the validator will be based on the HTML spec instead, and change its rules whenever the spec changes? That might be OK, as long as the spec changes are backwards compatible. If not, I suspect both the spec and the validator will become irrelevant.

This verbose blah blah is what we'll have to put up with . Errors, mistakes or too hight ambitions aside. One of the reasons i'm not p*issing my pants over HTML5.

So how are we going to check for HTML syntax errors in the future (I'm not being sarcastic)? Perhaps we could use an HTML4 Doctype while validating, and just ignore all the resulting errors about unknown HTML5 elements. Or maybe write our own custom DTD with all the HTML5 elements included?

[Darin McGrew]

So how are we going to check for HTML syntax errors in the future (I'm not being sarcastic)? Perhaps we could use an HTML4 Doctype while validating, and just ignore all the resulting errors about unknown HTML5 elements. Or maybe write our own custom DTD with all the HTML5 elements included?

Yeah, good question... One of the things that stops me from doing much with HTML5 is that my process involves using an offline validator. I don't want to give up the offline validator, but downloading, installing, and configuring an HTML5 validator/checker is non-trivial.

[pandy]

So the validator will be based on the HTML spec instead, and change its rules whenever the spec changes?

the worst is that somone needs to do that. we can't just feed a dtd to a validator anymore and it will understand it. somone needs to interpret the spec for the validator. i guess that in reality means change the validator.

Or maybe write our own custom DTD with all the HTML5 elements included?

that's a brilliant idea! Liam! Where are you?

i have a problem with the verbosity of the 'new' (5-10 years?) w3c html validator too. i hate to read that blah blah every time. i like it short and succinct as it used to be and as 'our' validator still kindly expresses itself. i want to be able to sift through the errors quickly. i don't want to read a novel.

[pandy]

BTW, why was SGML abandoned? I don't think I've ever heard the reason for that.

[Christian J]

Come again?

[pandy]

SGML, the language DTDs are written in. Why is that not good enough all of a sudden? I guess my question really is why are DTDs bad?

[Darin McGrew]

Why is that not good enough all of a sudden? I guess my question really is why are DTDs bad?

I didn't find a concise (quotable) explanation, but my take on it is that SGML includes features that were never supported by browsers (although technically allowed by the DTD, because it was written in SGML), and that SGML cannot express features that are well-supported by browsers. So rather than try to fit HTML into the SGML box (or into the XML box), they decided to define HTML on its own.

But this is just conjecture based on reading various vague references to HTML5 and its lack of (official) DTDs.

[pandy]

you are probably right. i'll miss it anyway.