Re: Text hyphenation

[address@hidden]

On Sep 28, 2011, at 5:31 PM, Werner LEMBERG wrote:

It would be awesome if LilyPond had many of the functionalities of

LaTeX, but it'd be a pity if we spent developer time implementing a

host of LaTeX-like functionality from scratch when LaTeX already

exists.

Well, we don't want to imitate LaTeX, but naturally the same problem
appears, namely how to enter special characters conveniently.
Theoretically, none of the proposed commands are necessary since for
all of them you could use Unicode characters:

 \"  U+200B ZERO WIDTH SPACE (ZWSP)

 \|  U+200C ZERO WIDTH NON-JOINER (ZWNJ)

 \~  U+200D ZERO WIDTH JOINER (ZWJ)
     + U+0020 SPACE
     + U+200D ZERO WIDTH JOINER

True - my question was more oriented to LaTeX input syntax.  That is, if our goal is to use LaTeX(ish) syntax as much as possible, it'd be great if there were already open source parsers of LaTeX (or LaTeX like) text that could feed it into Pango or what have you.

My question still stands: are there any libraries that exist that

handle this sorta thing (i.e. is to typesetting text what pango is

to font)?

I think we should concentrate on using Pango, which also comes with
layout support (probably rather simple, I don't know).  If
improvements are necessary, we should ask the Pango team to provide
it.

http://developer.gnome.org/pango/stable/pango-Layout-Objects.html

http://developer.gnome.org/pango/stable/pango-Vertical-Text.html

For hyphenation, there exists the libhyphen library (used e.g. in
OpenOffice).

http://packages.debian.org/fr/sid/libhyphen-dev

luatex also allows paragraph building using its programmable lua
interface, but any dependency on TeX-like software means dependency on
TeXLive or an equivalent TeX distribution...

From the luatex website: "The sources may or may not compile, depending on the third digit after the comma of the local temperature in Dordrecht, Schwetzingen or Hasselt."

It seems like Pango and libhyphen are good ways to go.  A lot of our home-brewed algorithms can be retired for their more robust ones (unless the LilyPond text engine beats Pango, which is of course possible but I'd have to look into it).

Bertrand, as you're the one leading the charge for this, perhaps you could hold off until the coding par-tay!!1 ?  It'll save you the trouble of maintaining your own hyphenation package.

Cheers,

MS