On Wed, May 22, 2019 at 03:19:40PM -0400, LM wrote: > ... If you code your own toolkit, you can decide to handle only a few "unicode scripts". You may add a few limitations on top, and it can become way simpler to code. This is what I did with the drop-in replacement of harfbuzz: only basic left-to-right roman unicode scripts with pre-combined unicode code points. But it was ages ago, quite naive, and I could bet that the C API did change and broke my implementation (I used tinycc as a reference compiler :) ).
If I recall properly, the EFL did give the possibility to choose between a basic roman script engine or harfbuzz. GTK+ is hard dependent on harfbuzz though. freetype does not hard depend on harfbuzz, it's optional (I run such a freetype lib). I would recommend you to read on unicode grapheme cluster (the unicode "unit") and some "nasty" scripts like arabic/korean/thai to give you an idea on how far it can go and expect worse if you want to go global support. -- Sylvain