From: Philippe Verdy (verdy_p@wanadoo.fr)
Date: Fri Oct 28 2005 - 20:43:05 CST
From: "Kenneth Whistler" <kenw@sybase.com>
>> The rulings can be drawn and rendered using HTML only, without images,
>> but
>> there are lots of CSS tricks to do that, and the page is not accessible.
>
> It doesn't look like rocket science to me:
>
> <span style="border-top:1px solid; margin-top:1px;padding-top:1px;
> text-decoration:overline">
>
> The style definition of "text-decoration:overline" might not
> be accessible, but the in-line definition of the 2nd overline
> for the double-overline ought to be a clue.
Of course I know that. I personnaly wrote that code (note overline considers
only the case of 1 rule over digits; there's no style style to define 2
rulings or more avove text!). The code I used on the Wikipedia article is
more complex than your simple one:
This is tricky, because you have to play and add more boxes (where extra
<span>elements), change border styles, change border colors (this is the
only way to apply multiple rullings, as text-decoration applies only once
for the text content but not for surrounding boxes!), not being sure that it
matches the text color or does not conflict with a black background, and not
being sure that the extra boxes will fit on the line height...
All these tricks break the model of separation between text and style, and
break the compatibility with accessibility rules. To make sure that it will
display correctly, you have to force and fix all fonts, the text and
background colors, and all font sizes. But if a user still selects to not
apply any CSS rule, the text appears broken).
And that's the main issue: you can't automate easily the rendering by making
a macro that can be inserted anywhere in a text to reproduce these
supplementary digits. So this has to be carefully encoded specifically on
each page, and make sure that no page will be displayed with reversed
backgrounds (by users using visual accessibility features, or by attemping
to insert you own Latin text in other areas that have different composition
styles).
So the text is not easily reusable except on pages with very basic style
that you can guarantee will be always present as the default (such as black
text on white background, and no roman digits in indexable titles or plain
text searches or for building navigation anchors...)
Imagine what happens when I want to index the 100,000th century Before
Christ using Roman digits, for example in a TOC or general index, where
there are lots of centuries and all milleniums written with Roman digits.
Now you also want that the indexed title being searchable, viewable and
clickable, and easily composable for input: no other choice than using
European digits inconcistently, or you have to use some of the many
conventions used to write large roman numbers with only the existing few
basic Latin capital letters.
Now suppose you're working on a translation of an old medieval book that
contains lots of number instances with large values, such as astrolonomy,
mathematics, economy, tax collection, geographic and demographic metrics...
With the standard Roman numbers, you are limited at around 4999, or you have
to use complex notations. You cannot always use the encoded IDD or CCIDD
roman "digit" ligatures.In a translation you also need tokeep the notation
with stretched M above or with multiple overlines (the value will be either
in the body of the text, or in a footer, but you have to keep the original
notation).
This archive was generated by hypermail 2.1.5 : Fri Oct 28 2005 - 20:45:09 CST