Re: Lout parsing

[Ted Harding]

( Re Message From: Ian Jackson )
> Blake McBride's proposal doesn't include unpredictable rules.  In
> order to ensure that a space is considered a middle-of-sentence space
> it is simply necessary not to break the line there; in order to ensure
> that it is it is simply necessary to add a linebreak or another
> space.
> 
> * Tom Gordon suggests requiring ~ for non-sentence-ending space.  I am
> strongly opposed to this.  TeX's requirement for using backslash-space
> in the same situation has confused and continues to confuse many new
> users, and makes the markup input look very ugly and hard to read.
> 
> There is no need for it, either.
> 
> Ian.

I've been following all this with interest. Clearly the non-white-space
characters (including "end-of-sentence" punctuation) do not disambiguate
sentence/mid-sentence space. The user will, at least sometimes, have to
intervene. The question is, what distribution of criteria between built-in
rules and user intervention will be most convenient?

As a hardened troff user, I have no difficulty in putting either a
new-line or 2 spaces (but I prefer NL) at the end-of-sentence following
"!", "?" or "."; on the other hand my editor may break lines as in "Dr.
James Joyce" creating pseudo end-of-sentence; but in any case I prefer to
avoid the possibility of it's printing out like as well; so I have trained
myself to type in "Dr.\~James Joyce" (the "\~" gives an unbreakable
stretchable space, as opposed to "\ " which is unstretchable).

Although this is a bit inconvenient and could confuse new users, the need
doesn't arise very often; while the saving of effort due to the bult-in
rule is enormous. Personally I find the troff rule the best of all
possibilities.

What would the discussants propose for the following sentence?

        If every "end-of-sentence" stop always meant end-of-sentence
        (Heaven forbid!) then we would have trouble with sentences like
        this.

(The point being that in troff the ")" is end-of-sentence transparent, so
that the "!" is recognisable as an end-of-sentence stop; but in the above
as typed the fact that it's in mid-line suppresses that interpretation.)

Best wishes,
Ted.                                    (address@hidden)