Re: lynx-dev Superscripts

[Klaus Weide]

On 6 Jun 2000, Sergei Pokrovsky wrote:

> Well, that's better than nothing, but I'd prefer a simpler TeX-like
> rule:
> 
> 1) normally <sub>foo</sub> should be rendered as ^{foo}
(I think you mean <sup>)

That's easy to change in the code (just compare the HTML_SUB logic
to the HTML_SUP logic, see reference to thread below).

> 2) exceptionally, if just one character appears in the superscript,
>    it may be rendered as ^f.
> 
> Point (2) is not important for me, actually some people may see it as
> a nuisance.

Point (2) wouldn't be trivial.  The code doesn't keep track of how
many characters appeared between <sup> and </sup> [and would that be
characters in the input stream, or in an intermediate (theoretical)
Unicode representation, or in the output rendering?  Example where
all three differ: "&reg;" -> U+00AE -> "(R)"].  Additional
complications arise if there are nested elements (including
<sup><sup>...</sup></sup>).

> I'm ready to tolerate "blah blah^{(TM)}.", it occurs very
> infrequently in the texts for which I use lynx.

Unfortunately I encounter such stuff (maybe more so with &reg;)
quite frequently; I assume that's true for most users who use
lynx in a general purpose way, without specific interest in math
or scientific pages.

> (I must admit that it is often used for references and similar
> attributions; but just in that case it is unpleasant that the
> reference becomes fused to the word it relates to.)
> 
> In W3 the rendering of such things is governed via style-sheets, and
> that seems to be the optimal approach.  Those interested in adequate
> rendition of the text can specify delimiters they prefer; the others
> may use colors of other typographical effect.  One can tune the
> rendition according to one's current needs.

I agree that style-sheets would be the way to control this.
Unfortunately the support for it just isn't there in lynx.

>   Klaus> The latter kind of use for SUP seems much more widespread
>   Klaus> than its use for math, for which support in HTML (and in
>   Klaus> lynx) is quite poor anyway.
> 
> I think the lynx audience is rather special one, which should rather
> rank the content higher than prettiness (though everybody wants the
> rendering to be as beautiful as it is reasonably possible).  E.g. I
> often want to use Lynx to dump a flat text.  No color effects, but I
> need to have some delimiters.

I'll leave that to the "lynx audience" then...  Personally, I think the
compromise I made is a good one, so I won't make a patch.  But if you
make one and get it included by Tom Dickey, I won't object.

>   Klaus> For your use of lynx (as indicated by your examples), you may
>   Klaus> want to change the source code.  Let me know if you need a
>   Klaus> pointer where to look.
> 
> Well, I am interested in such pointers.

This has come up before, so I just refer you to the archived messages:

   Linkname: Mail Thread Index
        URL: http://www.flora.org/lynx-dev/html/month032000/threads.html#00151

> But I want to remain within the standard.

There's nothing wrong with using a customized version of the code, if
that serves your purposes better.  That's the best currently available
equivalent to applying a user stylesheet.

   Klaus


; To UNSUBSCRIBE: Send "unsubscribe lynx-dev" to address@hidden