what kind of doctype is this?

On Mon, 26 Apr 2004, Sherm Pendley wrote:

D. Stussy wrote:
This sort of thing is always a good laugh. The fact that the "document
type declaration" is syntactically a COMMENT, i.e. it starts "<!",

Don't do much SGML, do you?

No, I don't.

An SGML declaration begins with "<!" and ends with ">". Inside of that,
comments are the stuff between matching sets of '--' delimiters.

Then I stand corrected. However, it in itself still isn't a pure HTML
construct, which is what one SHOULD have if it were part of the specification of
the language. Instead, it appears to be an escape to a different language.

* Thomas Lahn wrote in comp.infosystems.www.authoring.html:

(Note that not even HTML Tidy which is created by someone working for
the W3C includes the system identifier in tidied documents ...)

Oh, it does when it thinks it makes sense. For example if the input
document had a system identifier or if the output is XHTML. Always
adding one is asking for trouble with doctype-switch disabled browsers.

Wow, I am overwhelmed by all the advice on how to validate sites and
declare document types. I appreciate, most of all, the kind way people
tried to help in their posts... I have some serious reading to do! I just
bookmarked a bunch of links and will be reading through these.

Holly

Bjoern Hoehrmann wrote:

* Thomas Lahn wrote in comp.infosystems.www.authoring.html:
(Note that not even HTML Tidy which is created by someone working for
the W3C includes the system identifier in tidied documents ...)

Oh, it does when it thinks it makes sense. For example if the input
document had a system identifier or if the output is XHTML.

ACK

Always adding one is asking for trouble with doctype-switch disabled
browsers.

You must be kidding.


PointedEars

D. Stussy wrote:

On Mon, 26 Apr 2004, Sherm Pendley wrote:
D. Stussy wrote:
This sort of thing is always a good laugh. The fact that the
"document type declaration" is syntactically a COMMENT, i.e. it
starts "<!",

Don't do much SGML, do you?

No, I don't.

An SGML declaration begins with "<!" and ends with ">". Inside of that,
comments are the stuff between matching sets of '--' delimiters.

Then I stand corrected. However, it in itself still isn't a pure HTML
construct,

As pointed out before, even "<!-- ... -->" is not a "pure HTML construct".
You won't debate that SGML comments (/-- (?!--|--(?!--)*--)* --/ [CMIIW]
within /<! [^>]* >/) are extremely useful in HTML, will you?

which is what one SHOULD have if it were part of the specification of
the language.

Don't do much HTML either, do you?

<http://www.w3.org/TR/html4/struct/global.html#h-7.2>

Instead, it appears to be an escape to a different language.

HTML is a markup language that is defined by SGML markup.

,-<http://www.w3.org/TR/html4/>

[...]
HTML 4 is an SGML application conforming to International Standard
ISO 8879 -- Standard Generalized Markup Language [ISO8879].

There are both an SGML declaration of HTML:

<http://www.w3.org/TR/html4/sgml/sgmldecl.html>

and HTML DTDs that are written in SGML:

<http://www.w3.org/TR/html4/sgml/dtd.html>
<http://www.w3.org/TR/html4/sgml/loosedtd.html>
<http://www.w3.org/TR/html4/sgml/framesetdtd.html>


HTH

PointedEars

Jukka K. Korpela wrote:

Thomas 'PointedEars' Lahn <PointedEars (AT) web (DOT) de> wrote:
So how can something that can be obtained in other ways
be *required*? It simply cannot.

They just did. The definition of a markup system can impose any rules
that its designers want. They can surely require conformance to one of
three DTDs _and_ require that one of three specific <!DOCTYPE>s be useed
_and_ that the first character on line 42 must be an "A". If they so
wish.

So *their* validator violates *their* specification.
Thanks for clarification ...


PointedEars

On Tue, 4 May 2004, Thomas 'PointedEars' Lahn wrote:

Jukka K. Korpela wrote:

They just did. The definition of a markup system can impose any rules
that its designers want.

Unless they attempt to impose rules which are mutually contradictory.
E.g to define HTML as an application of SGML, and then to make rules
which contradict the rules of SGML. (Let's not go looking for
specific examples - I'm just making a general remark there.)

They can surely require conformance to one of
three DTDs _and_ require that one of three specific <!DOCTYPE>s be useed
_and_ that the first character on line 42 must be an "A". If they so
wish.

I guess so.

So *their* validator violates *their* specification.

Does it? Remember, the term "validation" has a rather precise
technical meaning in an SGML context. It certainly does *not*
guarantee that it will point up all the issues on which a document
does not conform to an HTML specification.

Thanks for clarification ...

Sorry, my irony detector is in for maintenance at the mo.

David Håsäther wrote:

Thomas 'PointedEars' Lahn <PointedEars (AT) web (DOT) de> wrote:
Firstly, you are wrong that "well-formedness" is independent of
the DTD. Invalid markup (i.e. markup that did not validate against
a DTD) does not fulfill the requirement of well-formedness, too.
Not necessarily.
-v please

(Actually, I don't know what "-v" means, [...]

Verbose. See for example
<http://www.trash.net/~reeler/iacd/doc/admin/node20.html>

Thanks for the example.


PointedEars

Alan J. Flavell wrote:

On Tue, 4 May 2004, Thomas 'PointedEars' Lahn wrote:
So [the W3C's] validator violates [the W3C's] specification.

Does it?

Unfortunately, it does, if we assume that the system identifier is required.

If the DOCTYPE declaration of the resource to be validated misses the system
identifier and there are no further errors, the document is called Valid
HTML (4.01). If we assume that section 19 of the HTML 4.01 Specification is
only informational (for which I fail to see proof), this is clearly a bug.

Remember, the term "validation" has a rather precise
technical meaning in an SGML context. It certainly does *not*
guarantee that it will point up all the issues on which a document
does not conform to an HTML specification.

If the DOCTYPE declaration is missing or uses an invalid public identifier
in the resource to be validated, then validation is impossible. So far, so
good. But if the user then chooses to select a DOCTYPE in the Validator's
form and revalidates the resource, the resulting DOCTYPE declaration does
not contain a system identifier (and the Validator calls the resource
"Tentatively Valid"). If we assume that the system identifier is required,
this is clearly a bug, too.

If we instead assume that the system identifier is _not_ required as section
19 of the HTML 4.01 Specification is not informational but instead either
contradictory or supplemental to subsection 7.2, neither of these two
behaviors can be considered a bug. Since I fail to see proof for the thesis
that section 19 is only informational, until further notice I will adhere to
the thesis that section 19 is supplemental to subsection 7.2 and thus a
*HTML 4.01* DOCTYPE declaration without system identifier is correct.


PointedEars

Thomas 'PointedEars' Lahn <PointedEars (AT) web (DOT) de> wrote:

Alan J. Flavell wrote:

On Tue, 4 May 2004, Thomas 'PointedEars' Lahn wrote:
So [the W3C's] validator violates [the W3C's] specification.

Does it?

Unfortunately, it does, if we assume that the system identifier is
required.

Our assumptions do not change the SGML rules. The HTML specification may
require a system identifier, but this additional requirement does not
affect _validity_.

If the DOCTYPE declaration of the resource to be validated misses the
system identifier and there are no further errors, the document is
called Valid HTML (4.01).

That's misleading, but the document is valid. Whether it is HTML (4.01)
depends on interpretation of the HTML spec.

until further notice I will adhere to the thesis that section 19 is
supplemental to subsection 7.2 and thus a *HTML 4.01* DOCTYPE
declaration without system identifier is correct.

Section 19 is confused and confusing, but the situation is pretty simple:
a document that otherwise conforms to the HTML specification but does not
have a system identifier in the DOCTYPE declaration is valid (a valid
SGML document) but it is not a conforming HTML document, so it is
incorrect to call it "valid HTML" or "valid HTML 4.01".

--
Yucca, http://www.cs.tut.fi/~jkorpela/
Pages about Web authoring: http://www.cs.tut.fi/~jkorpela/www.html