On Tue, 28 Dec 2004 19:13:21 +0100, Torsten Bronger
<bronger@xxxxxxxxxxxxxxxxxxxxx> wrote:
 
> * Treat every dot as an end-of-sentence unless it is immediately
>   followed by an <neos/> ("not end-of-sentence").
> 
> * Mark abbreviation dots, if followed by whitespace, with an
>   immediately following &#x200b; (zero width space).  [It would be
>   prettier to mark end-of-sentence dots this way, but this would be
>   much more invasive.]
> 
> * Mark abbreviations with <abbrev>e.g.</abbrev>.  The cleanest
>   solution, but in my special case *much* more difficult to
>   implement than the other two, because I have an input stream to
>   convert to XML, and when I see the dot it's already to late for
>   inserting a tag.

Find full-stops followed by spaces followed by a capital letter.
Admittedly, it won't find 'em all, but it might let most abbrev. cases
fall through unless they're somehow followed by a capitalisation.
Maybe the 'sentence' should be an accepted or even implicit inline fo
unit (at about the same level as the implicit inline glyph-stacking
line-building and line-breaking stage within a block)? It could have a
real-world use - imagine a user-agent that wants to offer the reader a
means to quickly 'skip' from sentence to sentence for some
focus-related reason?

-- 
Ian K Tindale
http://tindale.dyn.nu/