Re: CENTER element [Was: Netscape & New HTML]

Tony Sanders (
Tue, 25 Oct 1994 15:10:07 +0100

> > and <P ALIGN=CENTER> makes little sense
> > given that (a) standard practice is that <P> is end of paragraph, not
> > a container,
My, how quickly history is rewritten; <P> has never marked the end of a
paragraph. The rule was that <P>'s and </P>'s should be implied where
they are obvious. That was probably a bad idea but that has no bearing
on this issue.

So the text:
some text
more text<P>
the end
should translate to:
<P>some text</P>
<P>more text</P>
<P>the end</P>

> of a document, not just individual units. The proposed
> alignment attributes for paragraphs, headings and other
You can't just put <CENTER>...</CENTER> around arbitrary sections.
See the various SGML discussions about marking sections of text.

Specifically, this is highly illegal and problematical:

> This element may be added to the DTD with minimal changes:
Not without serious restrictions on it's use.

> There. What was so fucking hard about that?
The fact that it will not work as presented puts a pretty good damper on
it as a serious proposal. I suppose that is what is "so hard" about it :-)

> See how responsive MCOM is to the needs of the users?
> They wanted a <CENTER> tag and MCOM gave it to them.
There are 10's if not 100's of important presentation issues totally
bypassed by taking this approach and adding more and more kludges will
just make things unmanagable. Ever heard of creaping featursim? Well
guess what...

> If it were up to all the SGML purists on www-talk
> nobody would have any control over how their documents
> look at all.
I'm *not* an SGML purist by any stretch of the imagination (I don't even
know that much about it really) but it's clear to me that the proposals
like <CENTER> aren't solutions to the real problems.

If we are going to add presentation cues inside HTML then I propose
that we do it in an organized manor. Not by adding random tags
as we see fit.

Also, why hasn't someone implemented stylesheets? The groundwork has been
done already; it's not like it's an unsolved problem or anything. The
code isn't that complex (even the CERN linemode browser has some of the
basics already).