Lars Gullik Bjønnes wrote:
No. I am not sure... but it depends... a combining character can be used to produce accents as well... why not an umlaut on top of an grave on top of an 'e'.
The reason I suggest a unicode inset is that we already have it: the latex accent inset.
Of course you can start playing tricks with fancy underlying character types which are in fact composed of other things, but that is an astronaut design: it is overlayered, so many abstractions on top of abstractions, so many that you need as many complicated mechanisms to make it go fast - it is so far up in the sky that there is no oxygen left, and the brain stops to work.
I would think it's best to just start with getting what we already have to work in a basic unicode setting, and maybe extend to a few eastern languages if volunteers come and help out. Don't worry about composed Unicode glyphs for now - it's a corner case that can be handled once someone feels the heat (which will probably when hell freezes over AFAICT).
The big step that takes us 99.9% of the way is just going single-code-point Unicode.
Ligatures and other display headaches are handled by the toolkits these days, so don't loose sleep over those.
The trick is to make the job as small and simple as possible, and single-code-point unicode is a huge, monotoneous improvement over the implicit 8-bit encodings used now, so why make the job harder? There is always another release after the next one.
Regards, Asger
