Re: color: NCSA Mosaic, Netscape, and HTML3

Joe English (jenglish@crl.com)
Tue, 18 Jul 1995 17:53:08 -0700


Message-Id: <199507190053.AA13042@mail.crl.com>
To: www-html@www10.w3.org, www-style@www10.w3.org, montulli@strumpet.mcom.com
Subject: Re: color: NCSA Mosaic, Netscape, and HTML3 
In-Reply-To: <9507181243.ZM29294@strumpet.mcom.com> 
Date: Tue, 18 Jul 1995 17:53:08 -0700
From: Joe English <jenglish@crl.com>


Lou Montulli <montulli@strumpet.mcom.com> wrote:

> If multiple vendors are supporting these BODY tags then they
                                                ^^^^ -- "attributes"
> are obviously not vendor-specific.  Perhaps they are just a
> good idea that should be added to the HTML 3 spec?


Perhaps; perhaps not.  If you think they're good ideas, why not
submit a formal proposal to the HTML working group?  Heck, even an
informal proposal would do; Eric Bina just sorta casually asked about
background images and the BODY BACKGROUND attribute ended up in the
draft spec as a result.

BGCOLOR and TEXT sound like good ideas to me, except that TEXT 
should be named FGCOLOR.

My intuition says that the LINK, VLINK, and ALINK attributes are
bad ideas, since they presuppose a particular presentation model:

(1) Imagine a browser that displays body text in black, unvisited
links in red, and visited links in a shade of blue, fading from
light blue to black depending on when the link was last visited.
(2) Imagine a browser that does not use color to distinguish anchors
at all, but displays an icon in the margin instead.

Naturally, such browsers could and should just ignore the LINK,
VLINK, and ALINK attributes.  But if LINK, VLINK, and ALINK
belong in the spec, then so do LINKICONPLACEMENT (for the second
browser, to tell it whether to put the icon in the left
margin or the right) and VISITEDANCHORCOLORDECAYFUNCTION (for
the first browser, to tell it how to compute what color to use
as a function of how long ago the link was last seen.)

Naturaly, Netscape and Mosaic (and other browsers which use that
processing model) could and should just ignore the LINKICONPLACEMENT
and VISITEDANCHORCOLORDECAYFUNCTION attributes, but do these really
belong in HTML to begin with?  If they do not -- and I'm sure most
will agree that they don't, as no popular browser actually works that
way -- then neither do LINK, VLINK, or ALINK.

Plus, the attribute names are all wrong.

Then again, my intuition may be totally off.  (I thought 
the BACKGROUND attribute was a good idea too, but now that
it's been implemented and in widespread use, I've had to quit
using Netscape altogether -- I have a greyscale X terminal;  
'nuff said :-)


--Joe English

  jenglish@crl.com