> On 21 August 2018 at 01:04 "Mark E. Shoulson via Unicode" > <[email protected]> wrote: > > It is kind of a bummer, though, that you can't experiment (easily? or at > all?) in the PUA with scripts that have complex behavior, or even > not-so-complex behavior like accents & combining marks, or RTL direction > (here, also, am I speaking true? Is there a block of RTL PUA also? I guess > there's always RLO, but meh.) Still, maybe it doesn't really matter much: > your special-purpose font can treat any codepoint any way it likes, right? > > ~mark > > Back in 2006, I was typing the Tai Tham script (then being proposed as the Lanna script) using the PUA and exploring the issue of selecting between what are now <MEDIAL RA> and <SAKOT, RA> based on the preceding character and between what are now <SIGN AA> and <SIGN TALL AA> based on the preceding base character and its subscripts. I was also looking at using variation selectors to override the rules. I was using SIL Graphite fonts when they was getting intermittent support in OpenOffice and Firefox - my main display engine was WorldPad. Nowadays, SIL Graphite seems to be securely supported in LibreOffice and Firefox. Now, back then, Graphite was at least attempting to support RTL; I would expect the RTL support to work well by now.
On the other hand, experimenting with OpenType is much harder. The best I've found is transcoding to a Latin range and using an ssxx feature to convert the Latin glyphs back to those for the complex script. I do that to render Tai Tham in Internet Explorer 11 on Windows 7; this complex scheme is a fallback for when the rendering engine fails. Richard.
