Re: Fractions Behaving Badly in Browsers

On Tue, 28 Jun 2005, *** Margulis wrote:

> shamshoon wrote:
>
> > > Displaying the correct glyph is more important
> > > than getting the font style correct.

If the reader is upset by the font style, the solution would be to get
a font which contains all of the relevant characters. The browser
would not then need to resort to some other font in order to display
the glyph.

> > If the font style is incorrect and/or
> > impossible to control,

In a web situatation, the document author does not "control" the
presentation. The best that the document author can do is to specify
the required Unicode character, maybe suggest/propose a font, and rely
on the defined interworking specifications to get it displayed on the
reader's display (which might not be visual, by the way).

> > Sure, the correct glyph is important. But when's the last time you
> > worried whether "abcdefg" etc. was going to show up properly in a
> > browser?

There /is/ a difference between a plain us-ascii repertoire and a
comprehensive Unicode repertoire, though.

(And I'm certainly "worried" about what's going to come out if some
idiot specifies "W" to be displayed in the Symbol font! But I
digress...)

> The solution is to use piece fractions instead of case fractions.

Sorry, but this is no "solution". The document should contain the
appropriate Unicode characters (how else is a speaking browser
supposed to know how to read them out? How is an indexer supposed to
know what they are?). Building characters out of snippets can at best
be described as a kludge. The place to resolve this IMHO is in the
selection of a font which contains the required character repertoire.
Ultimately that's under the reader's control.

It may be noted that, of the fractions that are in U+21xx, the
"eighths" (U+215B to 215E) are in the MS-defined "WGL4" repertoire,
but the thirds, fifths and sixths are not:
http://www.microsoft.com/typography/otspec/WGL4.htm

> Build them yourself, in other words, using superscript numerator
> <sup>n</sup> followed by the fraction slash (&frasl; or &#8260; or
> &#x2044;) followed by a subscript denominator <sub>m</sub>. That way
> you stay in the font. Of course if a given browser on a given user's
> machine has trouble finding the fraction slash, I suppose all is for
> nought. But this ought to give you the more consistent results
> you're looking for.

The web has to be some kind of concordat between the author and the
reader. The interface between the two is the document that is sent
out from the web server to the reader, and that's covered by
specifications - to ensure, as far as practicable, that the document
will be usable for all of the purposes for which the web has been
defined. Visual display is only one of these purposes.

best regards
.