> From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf > Of Kenneth Whistler
> By the way, while Peter Constable noted that > > "the interaction of a boustrophedon with bidi is a valid issue." > > I would contend that that is at the higher level where bidi > interacts with the line layout mechanism that determines the > directional context, rather than inside the bidi algorithm itself... I was simply suggesting that there may be issues there to consider, but not necessarily implying that the bidi algorithm needed to address in any way interactions between boustrophedon and non-boustrophedon text. I *could not possibly* suggest that until I had some idea of what the expected behaviour should be. I'm not aware of there being any conventionally-defined behaviour for interaction between strong LTR or RTL text and boustrophedon text. The requirements have to be defined before a solution can be architected. And as you go on to point out... > Besides, this is really a very, very marginal concern. All real world > exemplars of boustrophedon are *not* bidirectional text, and all > real world exemplars of bidirectional text are not laid out in > boustrophedon. Why? Well, because it would be a stupid thing to > do and give readers and writers headaches. it's not obvious that requirements do exist. Peter Peter Constable Globalization Infrastructure and Font Technologies Microsoft Windows Division
