Re: external hyphenation

[Jeffrey Howard Kingston]

Well, I'm willing to write the Lout end of external hyphenation.  As
I see it, the logical design would be to place something in the .lh
file for the language concerned, namely the name of the other program
to call.  And also it would be logical for Lout to call this binary
once per paragraph, sending it a list of all the strings to be
hyphenated in that paragraph, one per line.

But I have some doubts about the advisability of the whole thing.
For one thing, starting up a binary once per paragraph sounds slow.
Also there is a technical problem with

    In Hungarian, when one doubles these sounds, only the first character
    is repeated. (eg. cch, not chch, or ssh, not shsh). The catch is that
    when hyphenating them, the lost character reappears, so ...cch...
    becomes ...ch-ch..., etc.

Irrespective of the method used to decide the hyphenation points, Lout
assumes that the result of hyphenation is a set of hyphenation points
inserted between the existing letters of the word.  There is no place
in Lout's paragraph breaking algorithm for local changes to other
letters near a hyphenation point.  So in fact if it asked for the
hyphenation points of

    metempsychosis

and got back

    met-empsychosis metem-psychosis metempsy-chosis metempsycho-sis

it would not know what to do with this information.  It would need
to get back

    met-em-psy-cho-sis

instead.

Jeff Kingston