The joys of multi-directional typesetting

Anyone who subscribes to the XeTeX or LuaTeX mailing lists will have seen a lot of posts from me over the past couple of weeks, pursuing various aspects of multi-directional typesetting. That might prompt some questions, not least what has raised this interest!

I got a bug report for siunitx a little while ago about interaction with the bidi package. That pushed me to look at something I’ve been wondering about for a while: what we need to do in the LaTeX3 codebase to allow for multiple directions (particularly right-to-left, which as bidi shows should be usable directly in LaTeX). It’s turned out that good documentation on the various primitives available is not so easy to come by, and that there is a lot to get your head around. Luckily, I’ve had some very useful feedback from a range of people, including John Plaice (one of the authors of Omega). It also turns out that not everything is quite settled in engine support for multiple directions, so asking questions is pretty important.

5 thoughts on “The joys of multi-directional typesetting

  1. What we are interested in is how people expect RTL to actually work. For example, we feel that we should describe paragraph shape in terms of the ‘start of lien’ and ‘end of line’ rather than ‘left’ and ‘right’. Would that make sense to RTL users?

  2. bidi is mainly of interest to me as it points out what’s expected/required by RTL users. The reason for looking at RTL stuff is to work out what needs to happen in the various piece of LaTeX3 code to allow for RTL ‘out of the box’, which is as much about concepts as it is about code.

Leave a Reply