Proposed refinements to HTML 3.0: motivations
Fri, 24 Nov 1995 01:31:20 +0100

I wish to propose some emendations to HTML Version 3.0.
Read this message only if you want. It follows a previous one containing only
the proposal, which are here repeated for ease of reference in the first part;
in the second part the motivations for those proposals are added.


1. I vote in favour of the proposed <C> (former <MARK>) tag.
2. Allow <T> and <BT> tags in ordinary text too (for example, inside
<EM>...</EM>, as in

Important: <EM>use an <T>upright</T> font here!</EM>

3. Allow WIDTH and ALIGN attributes for the HR element.
4. Introduce level 0 headings for the title of the whole document.

PART 2: MOTIVATIONS (87 lines)

I'm a novice in using HTML, so may be that my comments don't have much value.
Nevertheless, I want to express my opinion about a few points, which can be
divided into three categories: character-level markup, horizontal rules,
and headings.


1.a -- The <MARK> (or <C>?) tag

What happened to the proposed <MARK> tag? I've read that it has been
dropped "on advice of SGML open"; but in my opinion it is extremely useful,
and should absolutely be included in the standard. A new, shorter name has now
been proposed for this tag: <C>. I want to voice my opinion in its favour!

1.b -- <T> and <BT> in ordinary text too

There are situations where one wants to enforce the use of an upright
(roman) font in the midst of a phrase which is being rendered, say, in italics
because of a logical markup like <EM>. Example:

...see also <CITE>Derived Functors of <T>Hom</T>:
the <T>Ext</T> Functors</CITE>.

If the <CITE> logical style is mapped to italic physical style, the two words
"Hom" and "Ext" should be rendered in roman style; on the other hand, if the CITEd words are displayed in red or underlined, "Hom" and "Ext" should be red
or underlined too. Another tag, say <PLTXT>, could be used to reset ALL font
attributes (including colour) to their "normal" state... Is it too much?
One could observe that in the previous example "Hom" and "Ext" should
have actually been included in a MATH element, since they are operator names;
but there are other circumstances where this is absolutely inappropriate.
Suppose you want to cite a book whose title makes deliberate use of stylistic
variations to obtain a particular effect or to give itself a distinctive
appearance, something like

<CITE>The <I>Italic</I> <T>Roman</T>ce</CITE>

(a real-life example is the not-so-well-known italian magazine
<CITE><T>AVVEN</T><I>i</I><T>MENTI</T></CITE>, which uses the lowercase italic
"i" in the midst of the other uppercase roman letters as a distinctive mark).
In this case, the explicitly specified styles should take precedence over the
font style implied by the CITE element; but if citations are rendered, say,
in red, then the whole phrase "The Italic Romance" (or "AVVENiMENTI") should
be displayed in red too.
Of course, these problems could also be addressed by means of the <C>
tag mentioned above and the use of an associated style sheet, but this would
enforce the use of style sheets even in a document that wouldn't otherwise
need them. We could get rid of <I>, <B> and <TT> in the same way!
And finally, there is a reason of simmetry: if you allow <I>, <U> and
<B>, you are bound to allow <T> as well!


Netscape UA's understand ALIGN and WIDTH attributes for the HR element.
In general, I'm not favourable to automatically include Netscape extensions
to HTML in HTML version 3.0, but in this case I think it is better to do it.
According to present specifications, one has to resort to a style sheet if
he/she wants to draw a rule, say, from the left margin to the center of the
page, but not if he/she wants to enclose a DIVision containing text centered
between margins. I find it inconsistent to enforce the use of a style sheet
for an HTML-1.0 element like HR and not for an element like DIV that can only
be used by authors who must be aware of the existence of style sheets.
Moreover, almost all elements which are permitted within %Body.content have an
ALIGN attribute (as well as a CLEAR one).


--A heading level must be choosen in strict hierarichical order,
not on the basis of a particular visual effect.-- Perfect: I totally agree with
this statement. Thus, I always use a level 1 heading for the title of the whole
document, then several level 2 headings for the title of "chapters", and so on.
It's not much important, but this seems inelegant to me: I feel that a level
--zero-- heading should be used for the title of the whole document, level 1
headings for the title of chapters, and so on. A typical document should
include at most one level 0 heading (e.g., the title of a book), or no level 0
headings at all (e.g., if the document contains only a chapter of a book).

Thank you for your attention and patience.

Universita` degli Studi di Padova -- Italy